Raji Balasuubramaniyan, Senior Data Scientist, Manheim
summary
Manheim processes ~15 million cars a year and builds production models that segment dealers with K‑means into five groups (successful buyer, successful seller, buyer at risk, seller at risk, undecided), score churn with logistic regression and neural nets, add sentiment scores from Naive Bayes on buyer–seller messages, recommend inventory with user‑item and user‑user collaborative filtering plus a profit linear regression, and forecast activity with ARIMA using an 80/20 train/test split; outputs trigger sales and marketing interventions. 30/60/60+ day windows
AI-written summary · about
chapters (20)
- 0:00Introduction
- 0:37Agenda
- 1:12What is Manheim
- 3:15Churn rate
- 4:47ML approach
- 7:47Algorithm
- 9:47Logistic Regression
- 10:30Neural Network
- 10:48Sentiment Analysis
- 12:29Risk Undecided
- 13:47Recommendation
- 14:33Approach
- 16:57Clustering
- 17:20Collecting Data
- 19:06Sorting
- 20:14Forecasting
- 21:11Arima
- 21:44Forecast
- 22:17Summary
- 22:57Outro
related talks
08mdata science in automotive
01h 39mApplying Rigorous Machine Learning Methods in Business Strategy
27mA smooth ride: Online car buying and selling at mobile.de
25mApplied Data Science in decision making in Sales Department
29mAI-Driven Business Success: Measurable Results and Revolutionized Sales
11mAWS On Air ft. BMW & BCG
42mHow mobile.de brings Data Science to Production...
56mPredictive Analytics for Practitioners