Time to Calculate: How We Do Predictive Analytics at myTracker
summary
Describes how myTracker builds predictive analytics for games: discusses the forecasting pipeline, the specific challenges encountered (data and modeling issues), and the methods used to produce their predictions. Names myTracker and presents the work from a data scientist (Artem Khapkin) at the Games Gathering Kyiv 2020 conference.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 1:07What is myTracker
- 4:23What is LTV
- 8:04Mobile app revenue types
- 11:38Data for LTV prediction
- 14:52Predictive task formalization
- 16:20LTV prediction models
- 19:10Forecast for new users
- 20:36Forecast for apps with history
- 24:33Forecast for new apps
- 26:59Small-history apps forecast
- 28:45Prediction quality issues
- 33:30Model evaluation
- 38:37myTracker advantages
- 41:28myTracker limitations
- 43:13Q&A
- 49:22Summary
related talks
23mCustom-made games with machine learning and big data
56mPredictive Analytics for Practitioners
34mData Science Challenges in the Mobile Games Industry
19mHow to Leverage Predictive Analytics to Enhance iOS Marketing Performance
52mSQL Server Predictive Analytics: Customer Stories
04mInteractive Sports Analytics
41mFootball's data analytics in python
41mThe evolution of basketball with data science