Seeing at the Speed of Thought: Empowering Others Through Data Exploration

Slide 1

Slide 1 text

Seeing at the speed of thought Empowering others through data exploration Greg Goltsov Senior Data Engineer @gregoltsov www.gregory.goltsov.info (will have link to slides)

Slide 2

Slide 2 text

Seeing at the speed of thought Empowering others through data exploration

Slide 3

Slide 3 text

Seeing at the speed of thought Empowering others through data exploration

Slide 4

Slide 4 text

Seeing at the speed of thought Empowering others through data exploration yourself

Slide 5

Slide 5 text

Seeing at the speed of thought Empowering others through data exploration yourself your team

Slide 6

Slide 6 text

Seeing at the speed of thought Empowering others through data exploration yourself your team your company

Slide 7

Slide 7 text

Touch Surgery Built marketing/sales dashboards for Fortune 10 companies Built educational dashboards for 4 of the top 10 world-rated medical universities All from scratch

Slide 8

Slide 8 text

Appear Here World’s biggest online marketplace for retail spaces Internal recommendation system Highly visual debug interface for non-tech people

Slide 9

Slide 9 text

Southern Cross Austereo Modernising the data pipeline Spearheading data-driven culture throughout the company Datasets covering 80% Australians weekly

Slide 10

Slide 10 text

BI/DW tools

Slide 11

Slide 11 text

BI/DW tools

Slide 12

Slide 12 text

Remove barriers Make feedback fast Remove yourself

Slide 13

Slide 13 text

Remove barriers

Slide 14

Slide 14 text

Remove barriers Catalogued datasets with one-line import in Python Messy dataset in PDFs

Slide 15

Slide 15 text

Remove barriers Dashboard with right ﬁlters, Excel export “Can you run a query?”

Slide 16

Slide 16 text

Remove barriers. Foster curiosity.

Slide 17

Slide 17 text

Make feedback fast

Slide 18

Slide 18 text

Make feedback fast Found a new trend via tinkering “Tomorrow I’ll see results of the batch job”

Slide 19

Slide 19 text

Make feedback fast “Check the dash in 15 mins” “I put your request into the backlog”

Slide 20

Slide 20 text

Make feedback fast. Let people tinker.

Slide 21

Slide 21 text

Remove yourself

Slide 22

Slide 22 text

Remove yourself Data pipeline + products Ad-hoc

Slide 23

Slide 23 text

No content

Slide 24

Slide 24 text

No content

Slide 25

Slide 25 text

Remove yourself. Don’t stand in the way.

Slide 26

Slide 26 text

Remove barriers Make feedback fast Remove yourself

Slide 27

Slide 27 text

The goal is to turn data into information, and information into insight. – Carly Fiorina, former HP CEO

Slide 28

Slide 28 text

Insight Information Data

Slide 29

Slide 29 text

Insight Information Data Value ↑ Abundance

Slide 30

Slide 30 text

Insight Information Data Fraud Access pattern Logs

Slide 31

Slide 31 text

Insight Information Data Key inﬂuencers MOM trends Tweets

Slide 32

Slide 32 text

Ad-hoc queries Data pipeline Fast to develop Every query gets thrown away after Upfront investment Every integration builds foundations

Slide 33

Slide 33 text

Visualise your ETL. Augment your Data Warehouses with Data Lakes.

Slide 34

Slide 34 text

No content

Slide 35

Slide 35 text

Extract Transform Load Sources Data Warehouse

Slide 36

Slide 36 text

Extract Transform Load Sources Data Warehouse Data Insight Time

Slide 37

Slide 37 text

Volume Variety Velocity "3D Data Management: Controlling Data Volume, Velocity and Variety”, Gartner Inc. 2001

Slide 38

Slide 38 text

Volume Variety Velocity "3D Data Management: Controlling Data Volume, Velocity and Variety”, Gartner Inc. 2001

Slide 39

Slide 39 text

Analysis of Unstructured Data: Applications of Text Analytics and Sentiment Mining ~80% of all data is unstructured

Slide 40

Slide 40 text

~80% of your data is unstructured

Slide 41

Slide 41 text

http://www.ft.com/cms/s/0/de15414e-ebad-11e1-985a-00144feab49a.html#axzz2F3CM6G7g “Making sense of unstructured data isn’t about technology, it’s a business challenge”

Slide 42

Slide 42 text

Aberdeen Group research Don’t use unstructured data Use unstructured data Happy with the ability to share data 18% 60% Pleased with the accessibility 20% 50%

Slide 43

Slide 43 text

Volume Variety Velocity Machine learning "3D Data Management: Controlling Data Volume, Velocity and Variety”, Gartner Inc. 2001

Slide 44

Slide 44 text

Ingest quickly Real-time schema-on- read exploration Push vetted insights into DW/BI Example: Spark, AWS Athena, Microsoft’s PowerBI

Slide 45

Slide 45 text

Collect Store Process/ Analyse Sources Data Warehouse Data Insight Insight Time

Slide 46

Slide 46 text

Collect Store Process/ Analyse

Slide 47

Slide 47 text

Collect Store Process/ Analyse

Slide 48

Slide 48 text

Collect Store Process/ Analyse

Slide 49

Slide 49 text

No content

Slide 50

Slide 50 text

Look at data. A lot.

Slide 51

Slide 51 text

Look at data. A lot. http:/ /www.forbes.com/sites/gilpress/2016/03/23/data- preparation-most-time-consuming-least-enjoyable- data-science-task-survey-says

Slide 52

Slide 52 text

No content

Slide 53

Slide 53 text

No content

Slide 54

Slide 54 text

Scale computation and storage separately Go from non-trivial data to dashboard in minutes Spark is 20-100x faster than MapReduce Turnkey solution: www.databricks.com OSS: Apache Zeppelin on AWS EMR Spark

Slide 55

Slide 55 text

We made it! Now what?

Slide 56