0 - Introduction to analytics

Question 1

A (our) definition for analytics

Answer 1

Analytics is a process by which a team of people helps an organization make better
decisions (the objective) through the analysis of data (the activity)

Question 2

Categories of analytics

Answer 2

1) Descriptive analytics
-> What should we do?

2) Predictive analytics
-> What could happen?

3) Prescriptive analytics
-> What should we do?

Question 3

Descriptive analytics

Answer 3

The purpose of descriptive analytics is to reveal and summarize facts about what has
happened in the past or, in the case of real-time analysis, what is happening in the
present

This is done by examining, aggregating, and synthesizing data collected from a
variety of sources

Raw data are captured and recorded in source systems, eventually to be cleaned,
retrieved, and normalized such that entities and relationships can be meaningfully
understood

The audience for descriptive analytics is broad, potentially reaching all functions and
levels of an organization

Question 4

3 forms of descriptive analytics

Answer 4

1) Reporting

2) Visualization

3) Software

Question 5

Reporting

Answer 5

The real value of descriptive analytics comes from putting access to this plethora of
data into the hands of analysts who can use it to rapidly answer questions

Business metrics are decided (examples: inventory, workflow, sales, revenue), data
required are identified, data are collected and prepared, data are analyzed (summary
statistics), data is presented;
Standard reporting and dashboards
Ad-hoc reporting
Analysis/query/drill-down

Standard reporting and dashboards are useful to a point, but users need to be able to
“slice and dice” the data on the fly to gain more meaningful insights, computing
summary statistics and visualizing comparisons without being limited to predefined
reports

Question 6

Visualization

Answer 6

Descriptive analytics is often about communication, only limited math

It can be easily and quickly applied (simple calculations) in day-to-day operations

Question 7

Software

Answer 7

Most basic level: ordinary spreadsheets and databases

Systems designed specifically to support data visualization, exploration, and reporting
(Cognos, Tableau, and Spotfire)

These systems can greatly increase the accessibility of data and basic analytic
insights throughout an organization

Snapshot: It does not look beyond the “surface” of the data

Question 8

Predictive analytics

Answer 8

Predictive analytics seek to forecast the likely future state of the world (what will
happen?) through a deeper understanding of the relationships among data inputs
and outcomes

This is a much more demanding goal, so there is much more that can go wrong

Inexperienced analysts and leaders often imagine that once you have a good
descriptive model, you can use it to make good forecasts. Not true!

Statisticians have long understood that correlation does not imply causation

As a result, teams that wish to forecast the future need to use more sophisticated
modeling approaches and follow more rigorous validation procedures if they want to
have confidence that their forecasts make sense

Example: television programs that cover the stock market
-Every day, experts explain why the stock market behaved the way it did the previous day
-But can any of them accurately forecast what the market will do tomorrow? Not a one

Question 9

3 forms of predictive analytics

Answer 9

1) Data Mining and Pattern Recognition

2) Predictive Modeling, Simulation, and Forecasting

3) Leveraging Expertise