This project analyzes the performance of three different implementations (variants) of an in-app currency feature within a live video streaming service. The analysis focuses on daily revenue, user behavior, and retention rates to determine the best variant to roll out.
Variant 2 is recommended for roll-out due to its highest revenue per user and strong performance in the high purchase frequency segment. However, it is advised to incorporate insights from Variant 3, which shows significantly higher retention rates, to improve overall retention.
Data Preparation Descriptive Statistics and Visualization Statistical Testing Retention Rate Analysis Implementation Plan
Install Python 3.9.6 (if not already installed)
Follow instructions at: https://www.python.org/downloads/release/python-390/
pip install pandas matplotlib seaborn scipy lifelines scikit-posthocs Ensure you have the following packages installed:
- pandas
- matplotlib
- seaborn
- scipy
- lifelines
- scikit-posthocs
You can install these packages using pip:
pip install pandas matplotlib seaborn scipy lifelines scikit-posthocs
- Clone the Repository
- Prepare the Dataset:
- Ensure your dataset (abc_test_data.csv) is in the same directory as the script or update the file path in the script accordingly.
- Run the Analysis:
- python main.py
✅ Variant 2 is the most effective in terms of total purchased amounts.
✅ There is no significant difference in the single purchase values.
✅ Variant 2 is the most effective in terms of the number of purchases.
✅ Variant 3 shows significantly higher retention rates.
This README provides a comprehensive guide to running the analysis code, interpreting the results, and implementing the recommendations based on the findings. By following the steps and using the provided code, you can replicate the analysis and make data-driven decisions for feature roll-out.