A/B Test Analysis with Country Segmentation

Overview

This repository contains an A/B test analysis segmented by country to evaluate the effectiveness of a new landing page. The analysis includes data cleaning, merging datasets, exploring data distribution, calculating conversion rates, and performing statistical tests to determine if differences in conversion rates are significant.

Project Description

The goal of this project is to determine whether the new landing page (treatment group) performs significantly better than the old landing page (control group) across different countries. The analysis involves:

• Cleaning and merging datasets
• Exploring data distribution by country and group
• Calculating general conversion rates
• Computing conversion rates by country and group
• Performing statistical tests (z-test) to assess if differences in conversion rates are statistically significant

Data

The project uses two main datasets:

• ab_data.csv: Contains user conversion data including user ID, timestamp, group (control or treatment), landing page, and conversion status.
• countries.csv: Maps user IDs to their respective countries.

Hypotheses

• Null Hypothesis (H₀): There is no difference in conversion rates between the control and treatment groups.
• Alternative Hypothesis (H₁): There is a difference in conversion rates between the control and treatment groups.

A z-test was used to compare the conversion rates of the control and treatment groups, with a significance level (alpha) set at 0.05.

Analysis

Data Exploration and Cleaning

• Checked for null values, data types, and duplicate entries.
• Cleaned the data by removing duplicate user IDs to ensure each user is counted only once.
• Merged the cleaned datasets to combine conversion data with country information.

Data Exploration by Country and Group

• Analyzed the distribution of users by country and the number of users in each group (control and treatment).
• Calculated and visualized conversion rates for control and treatment groups by country.

Statistical Analysis

• Conducted z-tests to determine if differences in conversion rates between control and treatment groups are statistically significant.
• Calculated z-scores and p-values for each country.

Results

The z-test results show that none of the differences in conversion rates between the control and treatment groups were statistically significant at the 0.05 level. This indicates that any observed differences in conversion rates are likely due to random chance rather than a true effect of the new page.