1. Credit scoring in telecom (MegaFon)
Target: Identification of subscribers who are highly likely to return a loan or promised payment for the purposes of marketing campaigns.
Tools: Logit , regression analysis
2. Churn prediction, scoring of subscribers by the probability of churn in telecom (MegaFon)
Target: Identification of subscribers planning to switch operators to be retained through marketing campaigns.
Tools: Logit , Random Forest, Decision Tree, Regression analysis , SPSS modeler
3. Modeling as a part of researches in telecom industry (MegaFon)
Research objectives: subscriber segmentation, loyalty and satisfaction research, search for hidden behavioral factors, competitor analysis, brand health, effectiveness of advertising campaigns, media preferences , etc.
Tools: cluster analysis ( k – means , hierarchical cluster analysis), factor analysis (principal component analysis – PCA ), regression analysis, multivariate scaling , analysis of variance ( ANOVA ). For the period 2005 – 2014, more than 50 studies were conducted with samples from 150 to 3000 respondents.
4. Forecasting revenue and sales (MegaFon)
Target: Forecast of revenue and sales, identification of external factors affecting these indicators (seasonality, weather, days of the week, etc.)
Tools: Regression Analysis: Time Series
5. Model for predicting the gender of subscribers (MegaFon)
Target: Determine the gender of the subscriber in cases where it is not explicitly specified. This information is used to increase the reach of targeted marketing campaigns and as input to other models.
Tools: Logit , regression analysis, SPSS fashion designer
6. Modeling in the framework of a marketing research for the project ” The 5-Minute Economist “
Target: Based on the survey, determine the segments of potential readers of the book, identify explicit and hidden topics that are of interest to readers, insights.
Tools: cluster analysis (k- means), factor analysis (PCA)
7. Modeling within the framework of the project “The 5- Minute Economist”
Target: Determine the most and least economically developed states, countries. Reveal the relationship between the EPI index and other macroeconomic factors.
Tools: Regression Analysis, MapInfo
8. Subscriber churn maps, revenue maps, KPI maps (MegaFon)
Target: Building interpolated maps for various indicators based on base station data (thousands of stations, millions of subscribers, dozens of indicators: traffic, revenue, connections, etc.)
Tools: MapInfo, VBA in context of MapInfo