How granzoy/retentionanalysis works
Project type and purpose
• This is a data-analysis project (not a service or full-stack app). Its goal is to analyze Yandex.Rent revenue and retention via metric trees, unit-economics/cohort analysis, and an A/B test.
File structure
• Top level:
• AB testing.ipynb – Jupyter notebook for A/B-test analysis
• Unit economics.pdf/.xlsx – cohort and unit-economics data and analysis
• README.md – project description and usage instructions
• .gitattributes – Git config
• There is no “frontend/backend” split or microservice folder; this is a single‐pipeline data project.
README for additional insights
• Sections: Metric Tree, Unit Economics & Cohort Analysis, A/B Testing (Z-test for proportions, n=275 per group, 3-month duration).
• Tools: Excel, Python, statistical knowledge.
a. Main components
1. Data Sources
– Source files (.xlsx, raw CSV if any).
2. Data Preparation & Exploration
– Jupyter environment + Python libraries (pandas, numpy, scipy).
3. Metric Tree definition
– PDF artifact representing metric decomposition logic.
4. Unit Economics & Cohort Analysis
– Excel workbook + Python scripts/notebooks.
5. A/B Testing Module
– Jupyter notebook performing Z-tests, generating statistics.
6. Reporting / Visualization
– PDF exports (Metric Tree.pdf, Unit economics.pdf), in-notebook charts.
7. Version Control & Documentation
– Git/GitHub (README.md, .gitattributes).
b. Relationships and interactions
• Data Sources feed into Data Preparation.
• Prepared data flows into two parallel branches: Metric Tree & Unit Economics/Cohort Analysis.
• A/B Testing branch also uses prepared data and outputs statistical results.
• All branches converge into Reporting, producing final PDF artifacts.
• Version Control underlies the entire pipeline.
c. Architectural patterns/design principles
• Linear data-pipeline/workflow pattern.
• Modular analysis stages: each stage is a separate notebook or document.
• Artifacts (PDF, XLSX) as outputs of discrete steps.
d. Key technologies/frameworks
– Python, Jupyter Notebook
– pandas, numpy, scipy (Z-test)
– Excel
– PDF export (built-in notebook or external)
– Git (version control)