second-project-eda-inf-stats

Vanguard Digital Experiment Analysis

Overview

This project analyzes a digital experiment conducted by Vanguard to evaluate whether a new user interface (UI) improved process completion rates. The project involves data cleaning, exploratory data analysis (EDA), performance metric evaluation, hypothesis testing, and visualization using Tableau.

Team Members

• Esteban, Muskan, Carles & Ricardo
• Group 2

Project Structure

├── data
│   ├── df_final_demo.txt
│   ├── df_final_experiment_clients.txt
│   ├── df_final_web_data_pt_1.txt
│   ├── df_final_web_data_pt_2.txt
├── notebooks
│   ├── notebooks/cleaning_join_all.ipynb
│   ├── kpi_error_rate_eb.ipynb
│   ├── step_time_analysis_carles.ipynb
│   ├── kpi_conversion_funnel_eb.ipynb
├── visualizations
│   ├── tableau_dashboard
├── README.md

Datasets Used

• Client Profiles: Contains demographic information of clients.
• Digital Footprints: Tracks online interactions and engagement levels.
• Experiment Roster: Assigns users to the Control or Test group.

Methodology

1. Data Cleaning & Merging:

   • Removed duplicates and handled missing values.
   • Standardized data formats and merged datasets.

2. Exploratory Data Analysis (EDA):

   • Identified key demographic insights and engagement patterns.
   • Assessed initial differences between Control and Test groups.

3. Performance Metrics Evaluation:

   • Defined KPIs (Completion Rate, Time to Completion, Drop-off Rates).
   • Compared Control vs. Test groups using visual and statistical methods.

4. Hypothesis Testing:

   • Conducted statistical tests (T-tests, Chi-square) to determine significance.
   • Evaluated cost-effectiveness of UI changes.

5. Tableau Visualizations:

   • Created interactive dashboards to visualize trends and insights.

Key Findings

• The new UI led to a statistically significant increase in completion rates.
• Users in the Test group showed lower drop-off rates at key steps.
• The improved UI resulted in faster completion times.
• Cost-benefit analysis suggests rolling out the new UI across all users.

Tableau Dashboard

Check out our interactive Tableau dashboard for a deep dive into the experiment data.

Technologies Used

• Python: pandas, numpy, matplotlib, seaborn, scipy
• Jupyter Notebooks: Data exploration & analysis
• Tableau: Interactive data visualization
• GitHub: Version control & collaboration

How to Run the Analysis

1. Clone the repository:

   git clone https://github.com/estebanba/second-project-eda-inf-stats
   cd vanguard-digital-experiment

2. Install required dependencies:

   pip install -r requirements.txt

3. Run the Jupyter Notebooks for data analysis:

   jupyter notebook

4. Open the Tableau dashboard to explore visualizations.

Presentation link

https://docs.google.com/presentation/d/1hEOpsVXgPf42R0eSq3708uOtM84t6VEFCXYhEfr204w/edit?usp=sharing

Contributions

Feel free to contribute by submitting issues, feature requests, or pull requests.

License

This project is for educational purposes and is not affiliated with Vanguard.