How gabrielsmurillo/spacex-ibm_datascientist_capstone-project works
Project type and purpose
• This is a data-science/ML pipeline project with an interactive web dashboard.
• Purpose: Ingest SpaceX launch data, perform EDA and feature engineering, train and evaluate classification models, then expose results via a Plotly Dash app.
Main components
a. Data Ingestion
• External APIs (SpaceX REST API)
• Web scraping scripts (BeautifulSoup + requests)
• Local CSV / SQLite (spacex_launch_dash.csv, sqlite3)
b. Data Processing & Feature Store
• Jupyter notebooks for EDA, cleaning, feature engineering
• Intermediate data artifacts (cleaned CSV or SQLite tables)
c. Machine Learning
• Training notebooks (LogReg, SVM, Decision Tree, kNN)
• Hyperparameter tuning modules
• Model artifacts (pickled models or saved parameters)
• Evaluation metrics (accuracy, Jaccard, F1)
d. Dashboard Service
• spaceX_dash_app.py using Dash + Plotly
• Reads model predictions / data for visualization
• Dash server (Flask under the hood)
e. Reporting & Presentation
• LaTeX report (report/*.pdf)
• Presentation PDF (presentation/*.pdf)
Relationships & Data Flow
• From Processing → ML Training → Model Artifacts
• From Artifacts + Cleaned Data → Dashboard Service → End Users
• Reporting/PDF generation is a side flow drawing on processed data and evaluation results.
Architectural Patterns & Design Principles
• Pipeline pattern: clear stages (Ingest → Process → Train → Serve)
• Modular separation: notebooks/scripts each handle one concern
• Single-page interactive front end (Dash) communicating with a Python back end
• Reproducibility: static data artifacts (CSV/SQLite) and versioned notebooks
Technologies & Frameworks
• Python 3.8+, pandas, numpy, scikit-learn, BeautifulSoup, requests, sqlite3
• Jupyter for EDA and model development
• Dash & Plotly for the web UI
• LaTeX for the final report