Embed Size (px)
Citation preview
http://azure.microsoft.com/en-us/documentation/services/machine-learning/
http://azure.microsoft.com/en-us/documentation/services/machine-learning/
http://www.microsoftvirtualacademy.com/ebooks#9780735698178
http://www.oreilly.com/data/free/data-science-in-the-cloud.csp
Data Science and
Machine Learning
Essentials
With
Stephen Elston and
Cynthia Rudin
http://shop.or
eilly.com/prod
uct/06369200
40255.do
https://www.edx.org/
course/data-science-
machine-learning-
essentials-microsoft-
dat203x
http://www.amazon.com/
Predictive-Analytics-
Microsoft-Machine-
Learning/dp/1484212010/
ref=la_B00NBELJJI_1_1?s=
books&ie=UTF8&qid=14
41060294&sr=1-1
Module
Module
Col1, Col2, Col3, ………………………,ColN
Val11, Val12, Val13, ……………………,Val1N
ValM1, ValM2, ValM3, ………………, ValMN
………., ……….., ………., ……………………., ………
Rectangular table
N Columns - M Rows
Equal length columns
“Science is the systematic classification of experience.”
George Henry Lewes
Input data
Data Transformation
Train Model
Define Model Split Data
Score (prediction)
Evaluate Model
1925 2015Oldest cable
Even
ts L
ast
Year
0
20
f(x)=0 f(x)<0
f(x)>0
f(x) = function(
Events Last
Year, Oldest
Cable)
age>30age≤30
#classes>8 #classes≤8
Predict y=1
Predict y=-1
Predict y=-1
Predict y=1
Predict y=-1
Predict y=1
Predict y=-1
Predict y=1
FemaleMale
BschoolNo Bschool
laundryNo
laundry
P(U)
99.0%
P(D)
1.0%
P(D and -)
0.9%P(D and +)
0.1%
P(U and -)
4.95%P(U and +)
94.05%
P(+|D)
10.0%
P(-|D)
90.0%P(+|U)
95.0%
P(-|U)
5.0%
PredictedPositive
Predicted Negative
ActualPositive
TP FN
Actual Negative
FP TN
Predicted Positive Predicted Negative
Actual Positive TP FN
Actual Negative
FP TN
FPR
TPR
1
10
0
AUC
Random
Guessing
f (xi ) = b0 +b1xi
0 2000Number of Businessweek clicks
Inco
me
0
1,000K
f(x)
f (clicksi ) = 500K + 200 ´ clicksi
f(xi) = b0 + b1xi +
b2xi2 + b3xi
3
Azure ML Table R Device Port
myFrame <- maml.mapInputPort(1,2)
zip file
source("src/myScript.R")
maml.mapOutputPort(“myFrame")
print(“Hello world")
Azure ML Table Python Device Port
Def azureml_main(inFrame1, inFrame2)
zip file
import my_package
return “myFrame"
print(“Hello world")
https://gallery.azureml.net/Experiment/e616740e68c647ba9bbefa663d037df5
https://gallery.azureml.net/Experiment/c8c8fe15c4ee470685cc91d5e19c77dc
https://github.com/Quantia-Analytics/Contana-Data-Science-Example-R
https://github.com/Quantia-Analytics/Cortana-Data-Science-Example-Python
Col1 Col2
Val11 Val12
Val21 Val22 Azure ML
Service
Score
Out1
Out2
User
Application
Output
Input
Request
Response
End point
Score Trained Model
Output Transformation
Input Transformations
Published Input
Published Output
