💳 Ironhack Payments – Cohort Analysis Project

Data Science and Machine Learning Bootcamp – Ironhack Puerto Rico
📅 December 13, 2024 | 👩‍💻 Author: Ginosca Alejandro Dávila

---

🧠 Project Overview

This project analyzes transaction and fee data from a fintech platform using cohort analysis techniques. The goal is to uncover meaningful patterns in user behavior, retention, revenue, and incident trends, and translate those findings into actionable business insights.

Developed as a capstone project for the Ironhack DS/ML Bootcamp, the full workflow includes:

• Data cleaning & validation
• Exploratory Data Analysis (EDA)
• Cohort-based KPI computation:
  • Service usage frequency
  • Retention rates (monthly)
  • Incident rates
  • Revenue and cumulative revenue
  • ARPU (Average Revenue Per User)
  • CLV (Customer Lifetime Value)

📌 Note: Cohorts from 2019-11 and 2020-11 represent partial months and should be interpreted with caution.

---

💾 Project Structure

📂 project-1-ironhack-payments-2-en/ → Project root folder
├── 📂 project_datasets/ → Provided raw .csv and .xlsx files for analysis
├── 📂 cleaned_project_datasets/ → Cleaned and validated datasets exported from notebook 1 (.csv)
├── 📂 eda_outputs/ → Outputs generated during exploratory data analysis
│ ├── 📂 data/ → Aggregated tables and cohort-ready .csv files for later analysis
│ └── 📂 plots/ → Static .png visualizations generated from EDA
├── 📂 notebooks/ → Development notebooks for each phase of the project
│ ├── 📓 1_data_cleaning_ironhack_payments.ipynb
│ ├── 📓 2_eda_ironhack_payments.ipynb
│ ├── 📓 3_cohort_analysis_metrics.ipynb
│ ├── 📓 4_streamlit_app_dev.ipynb
│ ├── 📓 export_ironhack_payments_notebooks_to_py.ipynb
│ └── 📓 test_clean_scripts_colab.ipynb
├── 📂 scripts/ → Operational Python scripts auto-exported from notebooks
│ ├── 📂 annotated/ → Scripts with markdown headers and comments
│ └── 📂 clean/ → Production-ready scripts (e.g., 4_streamlit_app_dev.py)
├── 📂 reports/ → Final deliverables and written documentation
│ ├── 📝 1_data_quality_report_ironhack_payments.md
│ ├── 📝 2_eda_report_ironhack_payments.md
│ ├── 📝 3_cohort_analysis_report_ironhack_payments.md
│ ├── 📝 Ironhack_Payments_Cohort_Analysis_Presentation.pptx
│ └── 📝 Ironhack_Payments_Cohort_Analysis_Presentation.pdf
├── 📂 dashboard/ → Tableau dashboard assets
│ ├── 📊 ironhack_payments_dashboard.twbx
│ └── 📊 ironhack_payments_dashboard_screenshot.png
├── 📂 reference/ → Provided project instructions
│ └── 📘 ironhack_payments_project_instructions.md

---

📊 Dashboard Preview

🖼️

🔗 View Interactive Tableau Dashboard

Built using Tableau Public Desktop. The screenshot above shows a static preview of the full dashboard layout. The .twbx file is also included for those who wish to open the workbook locally.

---

🧪 Tools & Technologies

• Python (Pandas, Matplotlib, Seaborn)
• Google Colab – Main development environment for all notebooks
• Jupyter Notebook (local) – Used to test .py scripts for reproducibility
• Tableau Public Desktop – Dashboard creation and visualization
• Streamlit – Optional app interface (runs locally via Anaconda Prompt)
• GitHub – Project versioning and documentation

---

🚀 How to Run the Streamlit App (Locally)

Optional if you want to explore the app interface

1. Open your terminal or Anaconda Prompt
2. Navigate to the scripts/clean/ folder
3. Run the app:

streamlit run 4_streamlit_app_dev.py

This will launch the dashboard locally in your browser. The app summarizes cohort KPIs and visual trends interactively.

---

📈 Summary of Key Insights

• User Growth: Rapid user acquisition between April–July 2020, peaking in June
• Retention: Stronger mid-term retention in Feb–Apr 2020 cohorts
• Incidents: Surge in incident rates during May–July 2020, correlating with usage spikes
• Revenue: Highest revenue contributions in June–July 2020
• CLV/ARPU: Cohorts vary significantly in customer value, highlighting key opportunities for targeted retention strategies

---

🎯 Business Recommendations

• Prioritize re-engagement for cohorts with strong month-2 retention but quick drop-offs afterward
• Investigate operational issues during May–July 2020 to reduce incident rates
• Focus on onboarding practices used in Q1–Q2 2020 for future growth campaigns
• Consider targeted promotions for high-CLV user segments

---

🙋‍♀️ Author

Ginosca Alejandro Dávila
Data Scientist & Math Educator | Ironhack DS/ML Bootcamp Graduate
🔗 LinkedIn • GitHub