How anagonbu/proyecto_google works
Project Type and Purpose
• This is a data‐analysis & machine‐learning prototype project built around the “Medical Cost Personal Dataset” (a Kaggle CSV).
• The goal is to explore patient attributes (age, gender, BMI, children, smoking, region), build a predictive model for annual medical expenditure, and deliver visual/interactive reports (Tableau dashboard, R Markdown/PDF outputs).
File‐Structure Analysis
• Top‐level files:
• 00-insurance.csv → raw data source
• 01-EDA.ipynb → exploratory data analysis (Python notebook)
• 02_resume_model_…ipynb (2 variants) → modeling pipelines (Python notebooks)
• google2.Rmd → R Markdown report
• google2.html / .pdf → rendered report outputs
• Google_tableau.pdf → exported visualization/dashboard
• No “src/” or production “app/” directories, indicating this is an analytics prototype rather than a deployed service.
a. Data Layer
– Raw dataset (00-insurance.csv) stored in a flat‐file (or local CSV store).
– (Optional) SQL staging area for initial querying/preprocessing.
b. Processing & Modeling Layer
– EDA Notebook (01-EDA.ipynb): data cleaning, feature exploration, visualization code (Python + Seaborn).
– Modeling Notebooks (02_resume_model…ipynb): feature engineering, model training (scikit-learn, lazypredict), model evaluation.
c. Reporting & Visualization Layer
– R Markdown (google2.Rmd) → renders to HTML/PDF reports summarizing findings and metrics.
– Tableau Dashboard (Google_tableau.pdf and public link) for interactive visualizations.
d. Outputs & Deliverables
– Final deliverables: PDF presentations, HTML reports, Tableau interactive dashboard.
– Model performance metrics (R², MSE) and proposed campaign strategy.
Architectural Patterns & Design Principles
• Modular separation of concerns:
• Ingestion/EDA
• Modeling & evaluation
• Reporting & visualization
• Reproducibility: notebooks capture code & narrative; R Markdown ties code to report; version control via file naming.
Technologies & Frameworks
• Python ecosystem: Jupyter notebooks, pandas, Seaborn, scikit-learn, lazypredict
• R ecosystem: R Markdown (knitr)
• SQL (ad hoc queries)
• Tableau for interactive BI
• Output formats: HTML, PDF
Essential Elements Checklist
• [ ] Raw dataset component with path/name
• [ ] EDA process block
• [ ] Modeling pipeline block
• [ ] Reporting blocks (R Markdown, Tableau)
• [ ] Output artifacts (PDF, HTML, Dashboard link)
• [ ] Technology annotations (e.g., Python, R)