kalpanabaheti / gamesdau

GamesDAU

Summary of .py files -

1. dataloading: Loading CSV, viewing it via DataFrames, and extracting iOS and Android sub-datasets.

2. initialplotting: Methods for visualising the trends of multiple cohorts on a graph, and consequently visualising a full year of cohorts.

3. timeseriesanalysis: Methods developed for accurately identifying all significant junctures of change in DAU across cohorts

4. exceptionregionanalysis: Methods for understanding regions of ambiguity better using summary statistics.

5. exceptionregioninference: Assuming a linear curve fit (which can be switched to other approaches such as case-wise Bayesian perturbation, or other forms of regression, etc.), inferring a model to explain the exception region.

6. generalinference: Time-series based curve-fitting (exponential smoothing, to start with) on general trend of major alert regions per 60-day-runs across cohorts. This may also be used to extrapolate preliminary baseline for future data (predictive mode).

7. correlationDAU: Methods for assessing correlation between truncated previous summative DAU population and current cohort's Day 1 DAU.

Game Intelligence Official Report

Link: https://docs.google.com/presentation/d/10Ujh1dya4eUI_rUDGwhAjOecFQU2nhaHd1oHLCULJXI/edit?usp=sharing