This repository showcases a collection of A/B testing and experiment analysis projects built using R, Python, and SQL.
Each project explores how data-driven experimentation can inform product and marketing decisions through statistical testing, experiment design, and actionable insights.
To demonstrate practical experience with:
- Designing and analyzing A/B and multivariate experiments
- Applying statistical testing (t-tests, proportion tests, logistic regression)
- Using SQL for data extraction and aggregation
- Performing analysis and visualization in R and Python
- Communicating data-driven insights clearly and effectively
This project is licensed under the MIT License — feel free to explore, learn, and build upon it.
AB-testing-portfolio/
│
├── marketing-ab-test/ # End-to-end A/B test analysis (Python, R, SQL)
│ ├── data/ # Dataset used for analysis (CSV too large for GitHub preview)
│ ├── python/ # Python scripts, Jupyter notebooks, requirements.txt
│ ├── r/ # RMarkdown report, knitted HTML, R scripts
│ ├── sql/ # SQL query used in the analysis
│ ├── figures/ # Plots and visual outputs
│ └── README.md # Project-level summary and findings
│
├── LICENSE # Main repository license
│
└── README.md # Portfolio overview (this file)
Source: Marketing A/B Test Data
The dataset simulates user interactions across marketing campaigns, tracking exposure, conversions, and timing.
It is publicly available and used here solely for educational and demonstration purposes.
Languages: R, Python, SQL
R Packages: tidyverse, ggplot2, janitor, broom, dplyr
Python Libraries: pandas, numpy, matplotlib, seaborn, statsmodels
SQL: SQLite / MySQL syntax for querying and aggregation
- Experiment design and randomization checks
- A/A validation and power analysis
- Proportion tests and confidence intervals
- Logistic regression for uplift analysis
- Visualization of conversion metrics and lift
- Segment-level comparisons (device, region, campaign type)
Marketing A/B Test (Kaggle Dataset)
Analysis of marketing campaign variants to determine which ad performs best in driving conversions.
Includes data cleaning, randomization checks, proportion tests, logistic regression, and SQL queries.
- Add new experiments (e-commerce flow, user engagement)
- Compare frequentist and Bayesian testing methods
- Automate reporting dashboards with R Markdown or Streamlit
- Publish a short blog post summarizing findings
This portfolio was created by Danielle N. Cunes to demonstrate applied data science skills focused on experimentation, statistics, and actionable business insights, aligning with roles in Data Science, Product Analytics, and Experimentation.