This project analyzes the results of a digital A/B test conducted by Vanguard, a US-based investment management company. The experiment aimed to evaluate whether a redesigned user interface (UI) improved the process completion rate among clients.

Determine whether the new UI led to:

• Higher completion rates
• Statistically significant improvements
• Justifiable changes based on cost-effectiveness (≥ 5% improvement)

• df_final_experiment_clients.txt: Experiment group assignment (Test vs Control)
• df_final_demo.txt: Client demographic and account data
• df_final_web_data_pt_1.txt and df_final_web_data_pt_2.txt: Clickstream data detailing process steps

• Merged the three datasets on client_id
• Identified process completion by detecting the final process_step per client
• Filtered for valid gender values (M, F) and removed duplicates
• Created derived columns: completed, Variation, and demographic summaries

• Analyzed client age, gender, tenure, and interaction behavior
• Explored process completion patterns across groups
• Identified drop-off points in the digital funnel

1. Test vs Control Completion Rates (Z-test): Test group had significantly higher completion rate.
2. Cost-Effectiveness Threshold: Improvement exceeded 5%, justifying redesign costs.
3. Age Differences (T-test): No significant age difference between groups.
4. Gender Completion Differences (Chi-Square Test): No significant relationship between gender and completion.
5. Group-specific Gender Differences: Analyzed using Chi-Square tests within each group.

• Completion rate bar charts by group, age, and gender
• Funnel step drop-off visualization
• Tableau dashboards for interactive filtering and demographic insights

The data supports rolling out the new UI due to a statistically and practically significant improvement in completion rates. No demographic biases were detected, and further improvements are recommended using post-rollout behavioral feedback.

• Python (Pandas, NumPy, Matplotlib, Seaborn, SciPy)
• Jupyter Notebook
• Tableau
• Git & GitHub

• Project_Week5_cleaning.ipynb: Main analysis notebook
• README.md: Project overview and documentation
• Tableau file (.twbx) for dashboards
• Supporting data files (hosted externally via URL)

1. Clone this repository
2. Install required Python libraries
3. Open the notebook and run cells in sequence
4. Load the Tableau workbook to explore dashboards

© 2025 | Created for the Module 2 Data Analysis Project