Professional Documents
Culture Documents
Overview
Data Deluge and the Opportunity
An easy case study
Key Considerations in Data Analysis
Components of a Data Analysis Plan
Technology supported Decision Making
Data Deluge
We are now flooded with data !
Product Related Data:
Categories, Features, Usages
Production Related Data:
Product Mix, Fault Rate
Customer Related Data:
What does each customer require: Quantity, Features,
Quality Requirements
Employee Related Data:
Skill sets, Experience Levels, Role, Performance
Statutory Regulations
Intellectual Property related data
.......an exhaustive list is plain difficult
Natura pharma
needs your help
MoisturePlus sales
Sep
Oct
Nov
Dec
Jan
Feb
Gross
Sales
5280000
5501000
5469000
5480000
5533000
5545000
Target
Sales
5280000
5500000
5729000
5968000
6217000
6476000
Ad costs
1056000
950400
739200
528000
316800
316800
Web
Costs
105600
316800
528000
739200
739200
Unit
Price
1.9
1.9
1.9
Define
Break down problems
and data into smaller
pieces
Disassemble
Draw conclusions
Evaluate
Put everything back
and make
recommendations
Decide
CEOs reply
How much do you want to increase the sales?
Get back inline with target, we cant afford to miss.
How do we do it?
Well, thats your job ! .
The strategy is get people (mainly tween girls, age 11-15)
buy more
How much sales you think are feasible ? Are target figures
reasonable?
Most of customers have deep pockets, so theres no
practical limit
How are our competitors doing ?
No clear numbers, but I know they are 50-100% ahead in
moisturising product revenue
What is the deal with ads and web ads ?
We are shifting more ad budget towards Web
Disassemble
Break problem into manageable smaller problems
How do we increase sales?
Some Examples:
Oct
Nov
Dec
Jan
Feb
Gross
Sales
5280000
5501000
5469000
5480000
5533000
5545000
Target
5280000
5500000
5729000
5968000
6217000
6476000
Ad costs
1056000
950400
739200
528000
316800
316800
Web Costs 0
105600
316800
528000
739200
739200
Unit Price
1.9
1.9
1.9
If the report is right, the CEOs belief about tween girls must be
wrong
CEOs belief was the mental model you were using
Your mental model determines what you see from the world
Your statistical model depends on your mental model
Mental models should always account for what you dont know
(uncertainty)
If uncertainty is included in the model:
you will be in the lookout for ways to use data to fill gaps in
your knowledge and thus make better recommendations
What is to be done ?
mail CEO again to query about what s/he doesnt know on
MoisturePlus ?
Vendor
Lot
Size(units)
12/4/2012
20000
13/4/2012
PP General Wholesalers
21000
13/4/2012
Lovely Princess
17000
20/4/2012
RV Girl cosmetics
19500
27/4/2012
Pretty Me
14500
2/5/2012
21000
6/5/2012
Lovely Princess
19000
17/5/2012
PP General Wholesalers
20500
23/5/2012
RV Girl cosmetics
18300
From Data:
Most of the reseller names look like they sell products for girls
But, who are PP General Wholesalers ? Whom do they cater to ?
They shared data on their customers for the month of April 2012
Vendor
Percentage
ST Shaving supply
31
39
buyfrom.in
12
18
Business Intelligence
Integrate large pools of data from major enterprise data systems to data warehouses
Support decision making by enabling user to mine and extract useful information
Data mining: Can obtain types of information such as associations, sequences,
classifications, clusters, and forecast
The goal of business intelligence is: to provide individuals within the organization with
the right information, in the right format at the right time to facilitate faster, better
decisions.
BI : Multidimensionality
Measure
Pr
od
uc
t
Time
Revenue:
2,000,000.
Customer
Dimension
BI applications provide
Multi-dimensional
analysis - Means you
can analyze facts at
the intersection of any
combination of
dimensions:
Show me the Revenue
we made from
Customer A for Product
1 in January 2012?
KDD
SciPy, Jlab
Statistics/Scientific Computing
Spread Sheets
Summary
Be very clear about the analysis objectives : the
problem
Be very familiar with all aspects of what defines
your data
Develop and stay true to your data analysis plan
Be cognizant of which mathematical and software
tools can best solve your problem
Be thorough in your analyses, express openness to
additional investigations, yet be mindful of
limitations given the data and the tools you are
using
Thank You