Should Digital Analysts Become More Data Science-y?

Should Digital Analysts Become More Data Science-y?

Tim Wilson

#SPWK / @tgwilson

@tgwilson / #SPWK

Source: Flickr / KamiPhuc

October 10, 1993

@tgwilson / #SPWK

Mt. Katahdin

@tgwilson / #SPWK

Source: Google Maps

The end of a5-month journey

@tgwilson / #SPWK

Source: Google Maps

…which was a journey on foot.

@tgwilson / #SPWK

Source: Google Maps

@tgwilson / #SPWK

Which is like walkingfrom Barcelona to Moscow.

@tgwilson / #SPWK

Last year, I was on a different journey.

@tgwilson / #SPWK

Source: Google Maps

Digital AnalystFall 2015

Intermediate R UserEnd of 2016

These journeys were similar in many ways.

@tgwilson / #SPWK

Flickr / Ines Hegedus-Garcia

They were both challenging!

@tgwilson / #SPWK

They were both rewarding!

@tgwilson / #SPWK

They both turned out to have destinations that weren’t as clear as I thought.

@tgwilson / #SPWK

Last year’s journey

@tgwilson / #SPWK

Source: Google Maps

Digital AnalystFall 2015

Intermediate R UserEnd of 2016

Last year’s journey

@tgwilson / #SPWK

Source: Google Maps

Digital AnalystFall 2015

Data Science-y AnalystEnd of 2016

@tgwilson / #SPWK

But…

WHY?!

Because the Role of the Digital Analyst Is Evolving

@tgwilson / #SPWK

The Spectrum of “Analytics”

Digital Analyst

Marketer

Basic Metrics Segmentation Data Science

@tgwilson / #SPWK

Data Science?!Data Science

@tgwilson / #SPWK

Data Science

Wikipedia sayeth

employs techniques and theories drawn from many fields within the broad areas of:

@tgwilson / #SPWK

@tgwilson / #SPWK

@tgwilson / #SPWK

@tgwilson / #SPWK

is like boiling water.

PEAK1

Flickr / Revolution_Ferg

MSR Whisperlite

vs.

@tgwilson / #SPWK

@tgwilson / #SPWK

vs.

vs.

Programming Languages

Open Source (Free*)

Supported By Large Communities

Connectable to Digital Analytics(and Other!) Data Sources

* TANSTAAFL

@tgwilson / #SPWK

Clingmans DomeThe highest point on the Appalachian Trail…

Flickr / David Fulmer

…and where I was during my undergraduate commencement.

@tgwilson / #SPWK

Just as education arms us with the power of knowledge…

...“programming” with data is powerful.Flickr / David Fulmer

@tgwilson / #SPWK

Data Extraction

Flickr / Scott Swigart

@tgwilson / #SPWK

@tgwilson / #SPWK

One GA Account

Many Properties!

Many Views!

@tgwilson / #SPWK

22lines of code

@tgwilson / #SPWK

22lines of terribly written code

@tgwilson / #SPWK

17lines of terribly okay written code

38Google Analytics views queried

IN

19seconds

@tgwilson / #SPWK

@tgwilson / #SPWK

Pull all of the views I can access

Filter to the one account I care about

Filter to only the “production” views

Pull the data for each of those views

Add that data to my (filtered) list of views

Export a .csv

Pretty up the column names

@tgwilson / #SPWK

Yes!

Is this just an

alternative to Adobe’s

Data Warehouse?

…with more flexibility

…without the mystery delivery time

Reusabilityand

Extensibility

Flickr / Scott Swigart

@tgwilson / #SPWK

@tgwilson / #SPWK

Pull all of the views I can access

Filter to the one account I care about

Filter to only the “production” views

Pull the data for each of those views

Add that data to my (filtered) list of views

Export a .csv

Pretty up the column names

@tgwilson / #SPWK

Comparisons: (Top) 18 Web Sites

@tgwilson / #SPWK

Totals: Channel vs. Customer Segment

@tgwilson / #SPWK

Anomalies: Channels vs. Segments

@tgwilson / #SPWK

github.com/gilliganondata

@tgwilson / #SPWK

Data Science

@tgwilson / #SPWK

@tgwilson / #SPWK

“The clearest way into the universe is through a forest wilderness.”

- John Muir

“Numbers have an important story to tell. They rely on you to give thema clear and convincing voice.”

- Stephen Few

0

20

40

60

80

100

120

140

160

180

200

1/3 1/10 1/17 1/24 1/31 2/7 2/14 2/21 2/28 3/6 3/13 3/20 3/27 4/3 4/10 4/17 4/24 5/1 5/8 5/15 5/22 5/29 6/5 6/12

@tgwilson / #SPWK

Where does R really stand out?

Flickr / Marina del Castell @tgwilson / #SPWK

VS.

Variety!r-graph-gallery.com

@tgwilson / #SPWK

X X

Variety!r-graph-gallery.com

@tgwilson / #SPWK

Variety!r-graph-gallery.com

@tgwilson / #SPWK

How else does R stand out?

Flickr / Marina del Castell @tgwilson / #SPWK

My Perspective: Fall 2015

@tgwilson / #SPWK

Do the Analysis Communicate Results

@analyticshero

DATA VISUALIZATION

My Data Science-y Perspective

@tgwilson / #SPWK

Do the Analysis Communicate Results

@analyticshero

DATA VISUALIZATION

To be clear…

@tgwilson / #SPWK

…many tools offerpowerful data visualization.

Data scientists build visualizations that

are interactive and reactive.

@tgwilson / #SPWK

bit.ly/ga-explore

bit.ly/ga-explore

@tgwilson / #SPWK

bit.ly/ga-explore

@tgwilson / #SPWK

bit.ly/ga-explore

@tgwilson / #SPWK

bit.ly/ga-explore

@tgwilson / #SPWK

Data Science

@tgwilson / #SPWK

What is a p-value?

@tgwilson / #SPWK

“Informally, a p-value is the probability under a

specified statistical model that a statistical

summary of the data (for example, the sample

mean difference between two compared groups)

would be equal to or more extreme than its

observed value.”

– American Statistical Association

@tgwilson / #SPWK

@tgwilson / #SPWK

p-value:

Statistically significant pregnancy test.

This topic is…

Flickr / Howard Lake @tgwilson / #SPWK

As web analysts, how often do we consider:

Correlations?

Regressions?

Confidence Levels?

Confidence Intervals?

Outlier Detection/Removal?

Type 1 vs. Type 2 Errors?

Bayesian vs. Frequentist Approaches?

Flickr / Neil Piddock @tgwilson / #SPWK

@tgwilson / #SPWK

This is hard!

Flickr / Casey Fleser

Which Channels Have a Traffic Correlation?

@tgwilson / #SPWK

TODAY

The metric went down. Why?!

TODAY

Did the metric move enough for me to care?

Forecasts = The Future?

Actual ForecastTODAY

(NOT FOR THIS, AT LEAST)

Let’s pretend “today” is in the past!

Actual ForecastTODAY

PRETEND“TODAY”

Let’s pretend ”today” is in the past!

Actual Forecast

PRETEND“TODAY”

Actual Forecast

And let’s build a forecast from that point

PRETEND“TODAY”

We know a forecast won’t be perfect.

Prediction Interval Actual Forecast

PRETEND“TODAY”

Now we have meaningful context!

Prediction Interval Actual Forecast TODAY

What this can look like in the real world.

@tgwilson / #SPWK

And…drilling down and summarizing

@tgwilson / #SPWK

Mark Edmondson’s

Pre-/Post- Analysis Using a Bayesian Structural Time-Series Method

@tgwilson / #SPWKbit.ly/ga-effect

So, are YOU ready to put data science(-y-ness) on your career roadmap?

STATISTICS

@tgwilson / #SPWK

Source: Flickr / Xavi

DATA

VISUALIZATION

COMPUTER

PROGRAMMING

The Measure Slack Team

@tgwilson / #SPWK

http://join.measure.chat

dartistics.com

@tgwilson / #SPWK

R and Statistics for the Digital AnalystColumbus, Ohio, US – June 13-15, 2017bit.ly/r-stats-training

Introduction to RDAY 1The BasicsTaught by Tim Wilson and Mark Edmondson

Statistics for the Digital Analyst

DAY 2

Putting R into Practice

DAY 3

Real-World Digital Analytics ExamplesTaught by Dr. Michael Levin, Otterbein University

Building Those Real-World Examples (and Advanced Topics)Taught by Mark Edmondson and Tim Wilson

Flickr / Jeffrey Stylos

Slack/Twitter: @tgwilson

linkedin.com/in/tgwilson

[email protected]

Podcast: analyticshour.io

@tgwilson / #SPWK

2-D Tool: bit.ly/ga-explore

Website: dartistics.com

Slack: join.measure.chat

Training: bit.ly/r-stats-training