Investor Presentaiton
Predictive Use Cases
Aavas
FINANCIERS LTD
SAPNE AAPKE, SAATH HAMAARA
APPLICATION SCORE
LOGISTIC REGRESSION BASED PREDICTIVE MODEL
A model that predicts future performance of applications being processed. Custom model built for Aavas, tested on past data, off-us and on-us, rank
orders leading & lagging indicators
Increase approval and Optimize TAT without increasing risk
Risk separation: 10x between least and highest risk deciles
Improvement in segment approval rates
Acquisition - Entry
Barrier, Risk Based
Pricing
Underwriting -
Prioritization, Reject
Review
CHURN PREDICTION
Identify customers who will request BT, months ahead of time
Proactive retention tool, allow business early warning indicators to improve retention for good customers
From
XGB MACHINE LEARNING MODEL
Reactive Retention Process
To
Proactive Retention Process
Acquisition
Servicing
Closure Request
Retention
42
22View entire presentation