View
162
Download
0
Category
Tags:
Preview:
DESCRIPTION
Understanding when a customer would re-enter market for purchase can be critical for success of any marketing campaign. Using survival analysis business can identify customers who are most likely to re-enter market in specific duration i.e 1 months, 6 months etc.
Citation preview
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
In Market Timing – Vehicle Purchase
Analytics Consulting
Technology Consulting
Business Intelligence
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Objective : Identify In-Mark Timing for Customers
In Market Timing – Vehicle Purchase (Passenger Segment)
Identifying the time when customers would reenter the marketplace so as to
develop effective marketing campaigns.
Target marketing coupled with right timing when the customer will be in the
market "looking around" for the right deal.
Predict the time to next car purchase a function of customer demographics,
geographic, behavioral patterns, service oriented, societal factors, current
vehicle and or last purchase date
Solution: Survival Modeling approach for addressing the in-market timing
Identifying the occurrence and timing of a given event (Next Purchase).
Estimates the probability distribution that the customer will look for a new
vehicle within the next time period t (called survival time), given that the last
purchase took place t units of time ago.
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Benefits : Survival Modeling
Survival Modeling predicts the timing of customer in the market.
It also indicates how various factors such as Mileage, Year of Birth, Income, Battery
Replacement, Total Problem Count, and others impact the occurrence and timing of
customer coming back to market.
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Overall Methodology
The Survival model building process follows the following approach and steps:
Owner data base consists of two types of customers, Sample customers from both
types
Customers who have made single purchase and
Customers who have made multiple purchases.
We divide the entire dataset into development set and validation set (70/30)
Derive graphical representation of interaction of various variables (customer
demographics, transactions, service data, vehicle systems) with time to re-purchase
Enables to realize the overall shape of survival & hazard function and also
whether the variable significantly influences the probability to re-purchase or
not
Impact on re-purchase probability for different values of the same variable
With the reduced set of variables and knowledge of hazard and survival function
curve shape, build a model which gives probability of every customer to re-purchase
in a particular month and re-purchase in the time up to any particular month
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Modelling Phases
Sampling
Variable reduction
Survival and Hazard
Function
Estimating Customer In-Market Timing
Explanatory Data Analysis
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Sampling - Variable Creation
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Sampling - Variable Creation (Cont.)
Sampling Strategy
Survival time - The time between the last event in the history period and the time until the first event
in the target period. For censored cases, the last time at which customers were observed.
Choice - Choice variable is used to distinguish the censored cases (customers who have done next
purchase within the analysis period) from observed cases. Choice =1 for customers who has done
next purchase within the analysis period, else it is 0.
Defining the time periods
Analysis Period - The period from the starting time to the ending time ("today") is the experimental or
the analysis period
Target Period - Target period is used to define the choice variable. If a customer has bought again in
this period than the value of choice variable is 1.
History Period – History period is used to define the independent variables.
The "target" period used to define the choice variable and the "history" period used to define the
predictors, are not consecutive. As customers who buy a car will lookout for another car only after
a gap of some time.
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Estimating Customer In-Market Timing
• The final step is to estimate the probability of a customer to re-enter the market at a
given point of time.
• With each customer in the base being scored for predicted in-market probabilities for a
specified time (that is, number of months since the origin of time), each customer will
have P0 through P6 as its predicted in-market probabilities. P0 is the predicted in-
market probability at the beginning of month 1, which is the origin of time. P1 is the
predicted in-market probability at end of month 1; P2 is the predicted in-market
probability at end of month 2, and so on.
• Two sets of probability scores will be generated
• Number of customers by decile during specified months
• Number of customers by decile up to specified months
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Estimating Customer In-Market Timing
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Model Lift
www.valiancesolutions.com Email: analytics@valiancesolutions.com © 2014 Valiance Solutions
Valiance Solutions Private LimitedA-146, Opposite TCS building,Sector 63, Noida, U.P - 201306India.
Contact Person: Vikas KamraOffice No: +91 120 4119409Contact No: +91 8750068961
Recommended