Predict the future!
A Predictive Analytics
Primer
Thomas H. Davenport
Be a “Bhavisyavani” (The future predictor)
 What customers see or shop, all these data are collected and
analysed to predict what they will want in future from you.
Learning 1: Customer lifetime value (CLTV)
 How much customer will buy from the company over a time
 Do you have the next best thing?
 Do your website recommends the best product offering?
This questions are answered
by using Predictive Analysis.
Learning 2: Digital Marketing is at heart
Majority of the data are collected through
online customer touchpoints.
The manager needs to decide where will
be the maximum customers attracted and
data can be easily gathered.
Digital marketing strategies are important
in very useful for current as well as future
market developments.
Learning 3: Forecasting
Data gathered through
digital marketing is also
useful for production
planning.
Production department can
forecast demand with more
accuracy and provides
opportunity to fulfil
customer’s wish.
Manager Vs Data Scientist
A manager’s work is to
communicate data and
make meaningful
interpretation of analysis
presented by data science
team
To form hypothesis, gather
data and maintain properly,
a data scientist is working
around the area of statistics
and predictive analysis
modelling.
Statistics isn’t
wizardry!
 The myth in managers
that they don’t have to
crunch number or data
will speak for themselves.
 Managers have to
communicate right
meaning of right data at
right time.
Approach
to the
future
A Good Manager
Statistics
Assumptions
Data
The Data
What
customers
are
buying?
Most common barrier
for good predictive
analyse.
The Statistics Regression
Modelling
Analysis
Check
Validity
Set
hypothesis
Know which models to
use. Set a hypothesis.
Create a predictive
model.
The
Assumptions
Assumptions
in model
Check
validity
Modify
Re-run
modelling
Applicability
of model
A model without stated
assumptions is no
model at all!
The one clause that
makes model dynamic
with time
When any one of them fails, you fail!
Everyone assumed housing
price will go up
No one assumed
that it could come
down
When all succeeds, you succeed!
The Ultimate Checklist
Source of data
Data size and
characteristics
OutliersAssumptions
Conditions violation
assumptions
Data analysis is a spiral.
More you analyse, more
you drawn deep to the
core.
Abhishek Rana
Follow me on LinkedIn here..

A Predictive Analytics Primer

  • 1.
    Predict the future! APredictive Analytics Primer Thomas H. Davenport
  • 2.
    Be a “Bhavisyavani”(The future predictor)  What customers see or shop, all these data are collected and analysed to predict what they will want in future from you.
  • 3.
    Learning 1: Customerlifetime value (CLTV)  How much customer will buy from the company over a time  Do you have the next best thing?  Do your website recommends the best product offering? This questions are answered by using Predictive Analysis.
  • 4.
    Learning 2: DigitalMarketing is at heart Majority of the data are collected through online customer touchpoints. The manager needs to decide where will be the maximum customers attracted and data can be easily gathered. Digital marketing strategies are important in very useful for current as well as future market developments.
  • 5.
    Learning 3: Forecasting Datagathered through digital marketing is also useful for production planning. Production department can forecast demand with more accuracy and provides opportunity to fulfil customer’s wish.
  • 6.
    Manager Vs DataScientist A manager’s work is to communicate data and make meaningful interpretation of analysis presented by data science team To form hypothesis, gather data and maintain properly, a data scientist is working around the area of statistics and predictive analysis modelling.
  • 7.
    Statistics isn’t wizardry!  Themyth in managers that they don’t have to crunch number or data will speak for themselves.  Managers have to communicate right meaning of right data at right time.
  • 8.
    Approach to the future A GoodManager Statistics Assumptions Data
  • 9.
    The Data What customers are buying? Most commonbarrier for good predictive analyse.
  • 10.
    The Statistics Regression Modelling Analysis Check Validity Set hypothesis Knowwhich models to use. Set a hypothesis. Create a predictive model.
  • 11.
    The Assumptions Assumptions in model Check validity Modify Re-run modelling Applicability of model Amodel without stated assumptions is no model at all! The one clause that makes model dynamic with time
  • 12.
    When any oneof them fails, you fail! Everyone assumed housing price will go up No one assumed that it could come down
  • 13.
    When all succeeds,you succeed!
  • 14.
    The Ultimate Checklist Sourceof data Data size and characteristics OutliersAssumptions Conditions violation assumptions
  • 15.
    Data analysis isa spiral. More you analyse, more you drawn deep to the core. Abhishek Rana Follow me on LinkedIn here..