33
LondonR Deploying your Predictive Models as a Service via Domino API Endpoint s Jo-fai (Joe) Chow Data Scientist at Domino Data Lab 6/15/2015 1

Deploying your Predictive Models as a Service via Domino API Endpoints

Embed Size (px)

Citation preview

Page 1: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Deploying your Predictive Models as a Service via Domino API Endpoints

Jo-fai (Joe) Chow Data Scientist at Domino Data Lab

6152015 1

LondonR

Agendabull Background bull My Domino Experience

o Why

o How

bull Examples (Iris amp Stock Market)

bull Conclusions bull Q amp A

6152015 2

LondonR

I LondonR

6152015 3

All about my PhD project very interesting stuff hellip

LondonR

First Collaboration

6152015 4

httpblogdominoupcomusing-shy‐‑r-shy‐‑h2o-shy‐‑and-shy‐‑domino-shy‐‑for-shy‐‑a-shy‐‑kaggle-shy‐‑competition

LondonR

Recap what I really do

6152015 5

LondonR

Recap what I really do

6152015 6

Since last talk hellip

xgboost(hellip) h2odeeplearning(hellip)

LondonR

About Domino Data Lab

6152015 7

LondonR

Why I use Dominobull Data science is complicated

o Knowing how to fit a model is not enough o Variety of challenges from data analysis to production o There is no one-size-fits-all solution

bull I do not have timeskills for every single task bull I can use Domino to fill the gaps bull Focus on understanding problems improving models

and presenting results bull Speed up analysis in just a few clicks bull More time for family and other stuff

6152015 8

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 2: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Agendabull Background bull My Domino Experience

o Why

o How

bull Examples (Iris amp Stock Market)

bull Conclusions bull Q amp A

6152015 2

LondonR

I LondonR

6152015 3

All about my PhD project very interesting stuff hellip

LondonR

First Collaboration

6152015 4

httpblogdominoupcomusing-shy‐‑r-shy‐‑h2o-shy‐‑and-shy‐‑domino-shy‐‑for-shy‐‑a-shy‐‑kaggle-shy‐‑competition

LondonR

Recap what I really do

6152015 5

LondonR

Recap what I really do

6152015 6

Since last talk hellip

xgboost(hellip) h2odeeplearning(hellip)

LondonR

About Domino Data Lab

6152015 7

LondonR

Why I use Dominobull Data science is complicated

o Knowing how to fit a model is not enough o Variety of challenges from data analysis to production o There is no one-size-fits-all solution

bull I do not have timeskills for every single task bull I can use Domino to fill the gaps bull Focus on understanding problems improving models

and presenting results bull Speed up analysis in just a few clicks bull More time for family and other stuff

6152015 8

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 3: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

I LondonR

6152015 3

All about my PhD project very interesting stuff hellip

LondonR

First Collaboration

6152015 4

httpblogdominoupcomusing-shy‐‑r-shy‐‑h2o-shy‐‑and-shy‐‑domino-shy‐‑for-shy‐‑a-shy‐‑kaggle-shy‐‑competition

LondonR

Recap what I really do

6152015 5

LondonR

Recap what I really do

6152015 6

Since last talk hellip

xgboost(hellip) h2odeeplearning(hellip)

LondonR

About Domino Data Lab

6152015 7

LondonR

Why I use Dominobull Data science is complicated

o Knowing how to fit a model is not enough o Variety of challenges from data analysis to production o There is no one-size-fits-all solution

bull I do not have timeskills for every single task bull I can use Domino to fill the gaps bull Focus on understanding problems improving models

and presenting results bull Speed up analysis in just a few clicks bull More time for family and other stuff

6152015 8

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 4: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

First Collaboration

6152015 4

httpblogdominoupcomusing-shy‐‑r-shy‐‑h2o-shy‐‑and-shy‐‑domino-shy‐‑for-shy‐‑a-shy‐‑kaggle-shy‐‑competition

LondonR

Recap what I really do

6152015 5

LondonR

Recap what I really do

6152015 6

Since last talk hellip

xgboost(hellip) h2odeeplearning(hellip)

LondonR

About Domino Data Lab

6152015 7

LondonR

Why I use Dominobull Data science is complicated

o Knowing how to fit a model is not enough o Variety of challenges from data analysis to production o There is no one-size-fits-all solution

bull I do not have timeskills for every single task bull I can use Domino to fill the gaps bull Focus on understanding problems improving models

and presenting results bull Speed up analysis in just a few clicks bull More time for family and other stuff

6152015 8

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 5: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Recap what I really do

6152015 5

LondonR

Recap what I really do

6152015 6

Since last talk hellip

xgboost(hellip) h2odeeplearning(hellip)

LondonR

About Domino Data Lab

6152015 7

LondonR

Why I use Dominobull Data science is complicated

o Knowing how to fit a model is not enough o Variety of challenges from data analysis to production o There is no one-size-fits-all solution

bull I do not have timeskills for every single task bull I can use Domino to fill the gaps bull Focus on understanding problems improving models

and presenting results bull Speed up analysis in just a few clicks bull More time for family and other stuff

6152015 8

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 6: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Recap what I really do

6152015 6

Since last talk hellip

xgboost(hellip) h2odeeplearning(hellip)

LondonR

About Domino Data Lab

6152015 7

LondonR

Why I use Dominobull Data science is complicated

o Knowing how to fit a model is not enough o Variety of challenges from data analysis to production o There is no one-size-fits-all solution

bull I do not have timeskills for every single task bull I can use Domino to fill the gaps bull Focus on understanding problems improving models

and presenting results bull Speed up analysis in just a few clicks bull More time for family and other stuff

6152015 8

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 7: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

About Domino Data Lab

6152015 7

LondonR

Why I use Dominobull Data science is complicated

o Knowing how to fit a model is not enough o Variety of challenges from data analysis to production o There is no one-size-fits-all solution

bull I do not have timeskills for every single task bull I can use Domino to fill the gaps bull Focus on understanding problems improving models

and presenting results bull Speed up analysis in just a few clicks bull More time for family and other stuff

6152015 8

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 8: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Why I use Dominobull Data science is complicated

o Knowing how to fit a model is not enough o Variety of challenges from data analysis to production o There is no one-size-fits-all solution

bull I do not have timeskills for every single task bull I can use Domino to fill the gaps bull Focus on understanding problems improving models

and presenting results bull Speed up analysis in just a few clicks bull More time for family and other stuff

6152015 8

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 9: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

How I use Dominobull Interface

o Web or R

bull Examples o Hello World (Iris)

o Stock Market Forecast

bull Code Sharing bull Try it Yourself

6152015 9

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 10: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Web Interface

6152015 10

Control Panel

A List of Runs

Console

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 11: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Web Interface

6152015 11

Resource Usage (I found it very useful)

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 12: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

R Interface

6152015 12

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 13: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

R Interface

6152015 13

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 14: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

ldquoHello Worldrdquo Examplebull Classic dataset - Iris bull Four numeric features predictors (x)

o Sepal Length Sepal Width Petal Length and Petal Width

bull One categorical target (y) o Three species of Iris ndash Setosa Versicolor and Virginica

bull Using R to build a simple predictive model bull Saving the model for future use bull Deploying the model as web service bull Automatic version control

6152015 14

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 15: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Predictive Model

6152015 15

y = f( X1 X2 X3 X4)

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 16: Deploying your Predictive Models as a Service via Domino API Endpoints

httpsappdominoupcomjofaichowexample_iris

Upload and Run

6152015 16

Upload the R script to Domino (Web R)

Start the Run (Web R)

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 17: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Evaluate and Save

6152015 17

Print ldquoRandom Forestrdquo model summary

Model with highest 10-shy‐‑fold cross-shy‐‑ validation accuracy (ie best parameter setting)

Include statistics for future comparison

Finally save the model for future use

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 18: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Deploy

6152015 18

Model

This script 1) loads the model 2) takes four numeric inputs (X1 X2 X3 amp X4) and then 3) returns a prediction

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 19: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Deploy

6152015 19

Point to that script

Specify the function to call

Publish or unpublish the API

Domino automatically keeps all versions of your API

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 20: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

How to use the API

6152015 20

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 21: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Python API Example

6152015 21

X1 X2 X3 and X4

The four Iris features Sepal Length Sepal Width Petal Length and Petal Width

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 22: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Stock Market Forecastbull Historical stock data from Yahoo bull Using R to generate numeric features (x) bull Target (y) ndash Next Trading Day Change in Closing

Price bull Using R to build ensembles for forecast bull Configure scheduled runs bull Automatic version control bull API

6152015 22

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 23: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Predictive Model

6152015 23

Historical stock price data from Yahoo

x Multiple Technical Analysis Indicators

y Next Day Change in Closing Price

Predictive Model Ensemble of xgboost models For more info see appdominoupcomjofaichowexample_stock

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 24: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Scheduled Runs

6152015 24

Point to the R script

Schedule to run at a certain time every Weekday (more options available)

Re-shy‐‑publish API endpoint so it uses the latest results

Select different hardware tiers

Notify your friends colleagues clients

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 25: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Results Notification

6152015 25

Summary PDF

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 26: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Automatic Version Control

6152015 26

Latest Version One of the Previous Versions (I was experimenting with ggplot2)

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 27: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Stock API Endpoint

6152015 27

Stock Symbol (Ticker) for Query

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 28: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Code Sharing

6152015 28

Control Panel agrave Settings agrave One Click

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 29: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Try it Yourself

6152015 29

Register at wwwdominodatalabcom Help Quick Start Forum at supportdominodatalabcom

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 30: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Try it Yourself

6152015 30

Go to httpsappdominoupcomjofaichowexample_iris

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 31: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Set up your first API Endpoint in Minutes

6152015 31

Point it to your own projectInsert your own API key

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 32: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Conclusionsbull Data science is complicated bull Our time is important bull I can use Domino to save time bull It helps me to tackle some challenges

that are outside my comfort zone

6152015 32

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33

Page 33: Deploying your Predictive Models as a Service via Domino API Endpoints

LondonR

Thanksbull Mango Solutions bull My Colleagues at Domino bull More Info and Feedback

o jofaidominoupcom o Twitter matlabulous

o httpblogdominodatalabcom

bull Code o Iris Example ndash httpsappdominoupcomjofaichowexample_iris o Stock Example ndash httpsappdominoupcomjofaichow

example_stock

6152015 33