Online Casino Product Analytics

Project Description

The primary objective of this project is to provide product analytics for an online casino. It comprises three main components:

1. A compilation of the most commonly used Key Performance Indicators (KPIs) for measuring user engagement, financial performance, game performance, and the overall operational efficiency of an online casino.
2. SQL queries tailored to deliver specific product analytics for an online casino, aimed at addressing the following key questions:
   • The number of unique users and devices over a specified period.
   • The top 5 countries by the number of registered users.
   • The top 5 users by total payments within each of the top 5 countries.
   • The average session length per user across the entire website.
   • Daily Active Users (DAU) who have confirmed their email addresses, broken down by month.
   • The proportion of successful payment amounts from users with confirmed email addresses.
   • Statistical measures such as the mean, lower quartile, median, and upper quartile of the time intervals between successful payments for each user.
3. Estimation of A/B test performance, aimed at evaluating the effectiveness of changes implemented within the online casino platform.

Dataset Overview

The dataset comprises two main parts:

• Three relational tables stored in the ClickHouse DBMS:
  • event_log
  • user_info
  • payment_records
• An events.parquet file containing raw data on the A/B test.

Database Schema

Dataset on the A/B Test

The dataset comprises 835,357 records, each including the following fields:

• user_id: The user's identifier (object)
• user_group: The A/B test group (int64)
• time: The time of the user's action on the platform (datetime64)

Setup & Requirements

Software Requirements:

• Python version 3.x installed.
• Database Management System (DBMS): This project uses ClickHouse as its primary DBMS.
• Jupyter Notebook for running and sharing Python code, analysis, and findings.

Python Libraries:

• Pandas: For data manipulation and analysis. Install using pip install pandas.
• NumPy: For numerical operations. Install using pip install numpy.
• SciPy: For scientific computing, including the Chi-square test. Install using pip install scipy.
• SQLAlchemy: For SQL database connection and queries from within Python. Install using pip install SQLAlchemy.
• ClickHouse-driver: For connecting to ClickHouse DBMS from Python. Install using pip install clickhouse-driver.

Methodology

• Developing SQL queries tailored for the ClickHouse DBMS.
• Preliminary data exploration and analysis.
• Data wrangling using the pandas Python library.
• Application of the Chi-square test to estimate statistical significance.

Findings & Conclusions

A/B test performance evaluation: The test variant group showed a slightly higher conversion rate (25.0%) compared to the baseline control group (24.7%). This suggests the variant might be slightly more effective. However, the difference is not statistically significant. There is insufficient evidence to assert a meaningful difference in conversion rates between the two groups, underscoring the need for further testing or analysis to draw definitive conclusions.