73
Getting better and better at investigating and weaving digital data DATA STORIES Screen capture, Sherlock - BBC

Data Analysis Basics - Workshop (Frameworks)

Embed Size (px)

DESCRIPTION

This workshop was created to kick off free and open discussion among the Accounts Strategy, Social Media Analytics and Media teams of Nuworks Interactive Labs. It goes through basic frameworks and rules of thumb that may guide the analysis process. The flow does not cover specific statistical analysis directives -- it's meant to spark a discussion and an assessment of the way individuals look at data, in an easily understandable way. Slides containing data exclusive to employees were removed.

Citation preview

Page 1: Data Analysis Basics - Workshop (Frameworks)

Getting better and better at investigating and weaving digital data

DATA STORIES

Screen capture, Sherlock - BBC

Page 2: Data Analysis Basics - Workshop (Frameworks)

I. ContextII. Frameworks and Rules of ThumbIII. Describing and Analyzing QualitativelyIV. Visualizing DataV. Statistics Basics - Averages & CorrelationsVI. Business Language

Page 3: Data Analysis Basics - Workshop (Frameworks)

Why are we doing this?CONTEXT

Page 4: Data Analysis Basics - Workshop (Frameworks)

COMPANY

GROUP

Data-driven decisions

PERSONAL

Page 5: Data Analysis Basics - Workshop (Frameworks)

If you only want to guess how strong your ad is doing, e de sana nag-TV, radyo o billboard ka na lang.

Data-driven decisions

Page 6: Data Analysis Basics - Workshop (Frameworks)

COMPANY

GROUP

Data-driven decisions

Better Insighting & Efficiency

PERSONAL?

Page 7: Data Analysis Basics - Workshop (Frameworks)

Gather as much Wisdom (and Sales)

from data

OBJECTIVE

Screen capture, Sherlock - BBC

Page 8: Data Analysis Basics - Workshop (Frameworks)

Doing research isn’t a silver bullet.

Page 9: Data Analysis Basics - Workshop (Frameworks)

Rules of Thumb

LIST - CLUSTER - SORT

FIND THE STORY, FIRST.

MACRO MICRO

Page 10: Data Analysis Basics - Workshop (Frameworks)

I. ContextII. Frameworks and Rules of ThumbIII. Describing and Analyzing QualitativelyIV. Visualizing Data

Page 11: Data Analysis Basics - Workshop (Frameworks)

Framework

Page 12: Data Analysis Basics - Workshop (Frameworks)

Start with the broadest view.Before you zoom in.

Screen capture, Sherlock - BBC

Page 13: Data Analysis Basics - Workshop (Frameworks)

“Performance”

Page 14: Data Analysis Basics - Workshop (Frameworks)

When does it peak?HISTORICAL

Page 15: Data Analysis Basics - Workshop (Frameworks)

Did anything change?PREVIOUS

PERIOD

Page 16: Data Analysis Basics - Workshop (Frameworks)

Where did it go?COMPETITION

- HISTORY

Page 17: Data Analysis Basics - Workshop (Frameworks)

Anu-anong klaseng mga tao?

PROFILE

Page 18: Data Analysis Basics - Workshop (Frameworks)

Process

Page 19: Data Analysis Basics - Workshop (Frameworks)

Pre-work: Outline

Page 20: Data Analysis Basics - Workshop (Frameworks)

5 minutesYou want to save money.

How would you outline your analysis of your personal

expenses?

Page 21: Data Analysis Basics - Workshop (Frameworks)

ProcessList - Sort - Cluster - Track

Page 22: Data Analysis Basics - Workshop (Frameworks)

1. List

Page 23: Data Analysis Basics - Workshop (Frameworks)

get your data into a

spreadsheet

Page 24: Data Analysis Basics - Workshop (Frameworks)

DRUDGERYLincoln said something like,

"Give me six hours to chop down a tree and I will spend the first four sharpening the axe."Sarah Slobin, Visual Journalist, New York Times and Graphics Editor of the Wall Street Journal

Screen capture, Sherlock - BBC

Page 25: Data Analysis Basics - Workshop (Frameworks)

Gather ALL possible data important to your objective.

Marina Ekroos

Page 26: Data Analysis Basics - Workshop (Frameworks)

2. Sort

Page 27: Data Analysis Basics - Workshop (Frameworks)

rank everything

Page 28: Data Analysis Basics - Workshop (Frameworks)

Arrange as many attributes as you can.

TIMEPERFORMANCEPROPORTION

Till you start perceiving patterns.

Or a lack of them.

Page 29: Data Analysis Basics - Workshop (Frameworks)

Get acquainted.

STRENGTHWEAKNESS

NORMS

Page 30: Data Analysis Basics - Workshop (Frameworks)

3. Cluster

Page 31: Data Analysis Basics - Workshop (Frameworks)

classify common

elements

Page 32: Data Analysis Basics - Workshop (Frameworks)

Check what occurs together,

and what drives

HIGH-PERFORMERSMID-RANGE

LOW PERFORMERS

Page 33: Data Analysis Basics - Workshop (Frameworks)

Classify and list all relevant dimensions.

GENDERFORMAT

TYPESee leads into the

meaning behind the performance.

Page 34: Data Analysis Basics - Workshop (Frameworks)
Page 35: Data Analysis Basics - Workshop (Frameworks)

4. Track

Page 36: Data Analysis Basics - Workshop (Frameworks)

repeat steps 1-3

after some time

Page 37: Data Analysis Basics - Workshop (Frameworks)

WATCH DATA GROW UP.Data is about relationships.

A number doesn’t make sense by itself.- Me.

Page 38: Data Analysis Basics - Workshop (Frameworks)

Taking your temperature again and again.

BENCHMARKINGFind: What was different that made the data shift?

Page 39: Data Analysis Basics - Workshop (Frameworks)
Page 40: Data Analysis Basics - Workshop (Frameworks)

COLOR > SHAPEBrain Processing Speed

Page 41: Data Analysis Basics - Workshop (Frameworks)

Compare, not just time,but location,

types of users (demographics), etc.

FINDING MEANING

Screen capture, Sherlock - BBC

Page 42: Data Analysis Basics - Workshop (Frameworks)

15 minutesI’ll give you a dataset.

Tell the group what story you got from it.

Page 43: Data Analysis Basics - Workshop (Frameworks)

I. ContextII. Frameworks and Rules of ThumbIII. Describing and Analyzing QualitativelyIV. Visualizing Data

Page 44: Data Analysis Basics - Workshop (Frameworks)

DESCRIBING DATA

Page 45: Data Analysis Basics - Workshop (Frameworks)

HISTOGRAM vs BAR CHARTOnly Quantitative Categorical or Qualitative

Numerical

Num

eric

al

DimensionsN

umer

ical

Page 46: Data Analysis Basics - Workshop (Frameworks)

If your data looks like this...

What would you say about it?

0

1

2

3

4

12-15 16-19 20-24 25-29 30-34 35-39 40-44 45-49

Page 47: Data Analysis Basics - Workshop (Frameworks)

If your data looks like this...

What would you say about it?

0

1.75

3.5

5.25

7

12-15 16-19 20-24 25-29 30-34 35-39 40-44 45-49

Page 48: Data Analysis Basics - Workshop (Frameworks)

If your data looks like this...

What would you say about it?

0

1.25

2.5

3.75

5

12-15 16-19 20-24 25-29 30-34 35-39 40-44 45-49

Page 49: Data Analysis Basics - Workshop (Frameworks)

If your data looks like this...

What would you say about it?

0

1

2

3

4

12-15 16-19 20-24 25-29 30-34 35-39 40-44 45-49

Page 50: Data Analysis Basics - Workshop (Frameworks)

PROCESSING QUALITATIVE DATA

Screen capture, Sherlock - BBC

Page 51: Data Analysis Basics - Workshop (Frameworks)

“Codebook”

List - Sort - Cluster - Track

Page 52: Data Analysis Basics - Workshop (Frameworks)

F, 25-34, C2D- long for closeness-openness, yet are themselves afraid of disclosure (judgment, rejection, comparison) – relatively kimi, only one out of five gregarious (learned to defend mother from others)- married feel that they have learned to disclose and share more of themselves with their husbands; singles not yet at that stage – had to deal with judgment and “pagbabalewala” which hurts them the most – a kind of betrayal of trust- friends are for fun and forgetting problems – reminisce about happy times together, not really about trials gotten through – instead opt to go through these alone interactions must remain happy; value emotional endurance more than “opening up”- open up more through platforms (mas madali, mas hindi mahihiya, less possibility of seeing reactions/ dismissal) but also most afraid of back-biters

- music becomes escape – means to be happy and to relate to when late at night (nag-eemote)

- lifestyle/ talk shows much appreciated – of course, host authenticity is important, but also advice – want to know how to fix themselves – relationships and fashion, hobbies, tipid tips, completeness of details

- news- koreanovelas – spunky leads- game shows- musical variety- good market for unlitext – since prefer to

relate to barkada in detached manner – opt to send pahapyaw jokes and quotes (pagpaparamdam)

- also good for landlines – calling reserved for family and best friends, value telebabad time, cheap and easy

- also good for magazines – mostly Cosmopolitan and FHM – checking out other fashions, tips, kung bagay ba itsura nung mga babae; unlike 25-34 (gauging competition)

- group by paper (boring and outdated), communication (interactive – landline, mobile, internet – involving, young, uso), cable, TV – catering to everyone (middle class, middle aged)

- for social networking, counting friends (like affluent actives, and checking out where others are now, btu in less mean spirited way), 2 use multiply but more for (hi-res) pictures and shopping (value having online album), unlike affluent actives who don’t really like multiply because more complicated to upload – value ease of use

F, 25-34, C2D- wanting to uplift oneself – guilt because of feelings of having failed parents; wanting to assert one’s worth despite being alone, not having finished school, feeling that there’s something missing then throwing oneself into relationship, losing virginity - prone to religiosity- chona? – no real deep trials except pinapagsabihan ng asawa

-- gusto kong maging sikat vs. ayaw kong maging tanga vs. gusto ko kasama ako sa pinaguusapan, ginagawa ng mga kaibigan (left out)

- fear of investing oneself (betrayal of trust – through philandering and separation turning into weak social attachments) vs. fear of being worthless/ insufficient (having done the betraying themselves, or being left behind turning into controlling own aggressiveness) vs. fear of disclosure/ abandonment (not being listened to (binalewala) after trusting turning into relying on self to endure pains – no one else to rely on)

- more “practical”, proactive, goal-centric – even in platform use – group “simple life”: landline, TV, radio (basic, necessities, homebody); “luho”, pambata, busy feel, pero meron din ako para hindi mahuli – mobile, internet, movies, cable; oldies, pang-tambay, may nalalaman pero mas pampahinga lang/ pag walang magawa – newspaper, magazine, tabloid

- differences especially seen in mobile/ landline values: landline seen as more practical, whole family can use, and mahabaan, for emergency, way to reach everyone (relatively have fixed talking schedules, unlike affluent actives who prefer to be out and about instead of having a “routine”); cellular phones only for text and call when needed, don’t forward or message en masse, only 2 younger tried joining clans, one only when bored but doesn’t take phone everywhere – even do not like using phones for music or games, use separate gadgets

M, 25-34, C2D- wary of women who now cheat – say it hurts more, because cheating for women means lack of love, but all flirt with women, go out, say it’s natural – attribute women’s philandering to technology making it easy. They themselves use text for flirting, but are careful- Half only have landlines for internet, other than that, mostly for business to have stable number when working from home; others more dependent on cellular phones

F, 25-34 Tin-tin the girl really really close to her mom, lambingan with family (they gave me all that I asked for), have boyfriend in London – not really open to leaving family (wouldn’t know what she’s do without them)Tin-tin – single: wouldn’t know what she would do without her family, almost to the point of not really wanting to get married, mahilig manlambing, “lahat naman binigay nila sakin” Single: has older boyfriend who supplies everything for family (way to help family), angry at father, but not vengeful, instead remains quiet – just as mother did before she died, allows him to continue favoring younger sister even if she feels neglected Other single girl – father more ‘responsible”, stricter (come home on time, etc.), feels she has to take care of her mother, but generally privileged upbringing till recentlyMarried: all have “textmates” from time to time – just fun – all three have “tested” husbands through text; fairly committed mothers – two are working (one coming from broken family, does nto want same to happen, one got pregnant early and wants to discipline daughters to not do the same)

Page 53: Data Analysis Basics - Workshop (Frameworks)

Similar to what a word cloud makes you understand

But, I still want you to feel the process of it.

Page 54: Data Analysis Basics - Workshop (Frameworks)

“Codebook”

List of a good mix of responses.Cluster common answers.

Screen capture, Sherlock - BBC

Page 55: Data Analysis Basics - Workshop (Frameworks)

15 minutesAnalyze the comments you get

on your Facebook account.

Page 56: Data Analysis Basics - Workshop (Frameworks)

I. ContextII. Frameworks and Rules of ThumbIII. Describing and Analyzing QualitativelyIV. Visualizing Data

Page 57: Data Analysis Basics - Workshop (Frameworks)

BasicsVISUALIZATION

Screen capture, Sherlock - BBC

Page 58: Data Analysis Basics - Workshop (Frameworks)

MODELING“Craft”

Page 59: Data Analysis Basics - Workshop (Frameworks)

Formality

Upscale/ High-class

Mass market but formal, serious

High-class and highly formal tone

Mass market and casual, colloquial, friendly

High-class and casual, colloquial, friendly

Page 60: Data Analysis Basics - Workshop (Frameworks)

From what you know, what are the two main drivers of social media community success?

5 minutes

Map 5-10 different brands

Page 61: Data Analysis Basics - Workshop (Frameworks)

Factor 1 Factor 2

Brand 1 High High

Brand 2 High Medium

Brand 3 Medium Medium

Brand 4 Medium Low

Brand 5 Low Low

Page 62: Data Analysis Basics - Workshop (Frameworks)

Factor 1 Factor 2 Factor 3 Factor 4 Factor 5

Brand 1

Brand 2

Brand 3

Brand 4

Brand 5

Page 63: Data Analysis Basics - Workshop (Frameworks)

CHARTS:)

Page 64: Data Analysis Basics - Workshop (Frameworks)

PIEBARLINERADARSCATTER

SHARESTRENGTHGROWTH“SPREAD”POSITION

Page 65: Data Analysis Basics - Workshop (Frameworks)

0"

20000000"

40000000"

60000000"

80000000"

10000000"

12000000"

14000000"

16000000"

18000000"

Walmart"

Family"Guy"

Eminem

"Amazon

.com

"Target"

Lil"W

ayne

"Facebo

ok"

Samsung"M

obile"USA

"Subw

ay"

Drake"

Youtub

e"Starbu

cks"

Adam"Sandler"

Nicki"M

inaj"

Rihann

a"Ba

rack"Obama"

Katy"Perry"

South"Park"

The"Simpson

s"Michael"Ja

ckson"

Wiz"Khalifa"

Usher"

Will"Smith

"Macy's"

Lady"Gaga"

Taylor"SwiT"

Bob"Marley"

SkiUles"

Hou

se"

Beyonce"

CocaXCola"

Vin"Diesel"

Dr."Pepp

er"

Buffa

lo"W

ild"W

ings"

The"Hangover"

Disne

y"Oreo"

Spon

gebo

b"SquarePants"

Cand

y"Crush"Saga"

Kevin"Hart"

Megan"Fox"

Linkin"Park"

Jersey"Sho

re"

T.I."

The"Be

atles"

Farm

Ville"

Victoria's"Secret"

MiU"Rom

ney"

Reese's"

McD

onald's"

Global""Likers""

U.S."Page""Likers""

0" 50" 100" 150" 200"

Walmart"Family"Guy"

Eminem"Amazon.com"

Target"Lil"Wayne"Facebook"

Samsung"Mobile"USA"Subway"Drake"

Youtube"Starbucks"

Adam"Sandler"Nicki"Minaj"

Rihanna"Barack"Obama"

Katy"Perry"South"Park"

The"Simpsons"Michael"Jackson"

Wiz"Khalifa"Usher"

Will"Smith"Macy's"

Lady"Gaga"Taylor"SwiR"Bob"Marley"

SkiSles"House"

Beyonce"CocaVCola"Vin"Diesel"Dr."Pepper"

Buffalo"Wild"Wings"The"Hangover"

Disney"Oreo"

Spongebob"SquarePants"Candy"Crush"Saga"

Kevin"Hart"Megan"Fox"Linkin"Park"

Jersey"Shore"T.I."

The"Beatles"FarmVille"

Victoria's"Secret"MiS"Romney"

Reese's"McDonald's"

Millions'

Global""Likers""

U.S."Page""Likers""

Page 66: Data Analysis Basics - Workshop (Frameworks)

0" 5" 10" 15" 20" 25" 30" 35"

Walmart"Family"Guy"

Eminem"Amazon.com"

Target"Lil"Wayne"Facebook"

Samsung"Mobile"USA"Subway"Drake"

Youtube"Starbucks"

Adam"Sandler"Nicki"Minaj"

Rihanna"Barack"Obama"

Katy"Perry"South"Park"

The"Simpsons"Michael"Jackson"

Wiz"Khalifa"Usher"

Will"Smith"Macy's"

Lady"Gaga"Taylor"SwiS"Bob"Marley"

SkiTles"House"

Beyonce"CocaWCola"Vin"Diesel"Dr."Pepper"

Buffalo"Wild"Wings"The"Hangover"

Disney"Oreo"

Spongebob"SquarePants"Candy"Crush"Saga"

Kevin"Hart"Megan"Fox"Linkin"Park"

Jersey"Shore"T.I."

The"Beatles"FarmVille"

Victoria's"Secret"MiT"Romney"

Reese's"McDonald's"

Millions'

U.S.'Page'"Likers"'

U.S."Page""Likers""

What patterns do you see?

Page 67: Data Analysis Basics - Workshop (Frameworks)

What patterns do you see?

Page 68: Data Analysis Basics - Workshop (Frameworks)

BEGIN WITH THE END IN MIND.

What do you want them to remember?

Decide on your flow first.

Page 69: Data Analysis Basics - Workshop (Frameworks)

PATTERN. NARRATIVE. OUTLINE. Say what you’re going to say. Say it. Then, say what you said.

Donald Lim, former CEO of Yehey, Burger King and MRM

Page 70: Data Analysis Basics - Workshop (Frameworks)

Gain a better appreciation of what makes us special as Filipinos on social media, as opposed to the often-publicized Western statistics 1) size of following2) category – person/organization/program/device,etc.3) origin

Page 71: Data Analysis Basics - Workshop (Frameworks)
Page 72: Data Analysis Basics - Workshop (Frameworks)

Rules of Thumb

LIST - CLUSTER - SORT

DESCRIBE YOUR DATA HONESTLY- LET’S GROW OUR CODEBOOKS

MACRO MICRO

VISUALIZATION- FLOW FIRST, BEFORE CHARTS: WHAT’S YOUR STORY?- GROWTH - LINE OR AREA CHARTS; “DRIVERS”: MODELS - HIGH-CONTRAST COLORS: EMPHASIS

Page 73: Data Analysis Basics - Workshop (Frameworks)

Good luck with creating your own data narratives!Let the drudgery begin.

Screen capture, Sherlock - BBC