How vikranth3140/movie-revenue-prediction works
Project type and purpose
• This is a full-stack Machine Learning application for predicting movie revenues.
• Purpose: Ingest movie datasets, perform preprocessing & feature engineering, train multiple regression models, and expose predictions via (a) a CLI tool (main.py) and (b) a Streamlit web interface (streamlit_app.py).
File structure
• Top-level scripts:
• main.py – command-line interface for real-time predictions
• streamlit_app.py – web front-end using Streamlit
• “models” directory: each file implements one scikit-learn or XGBoost model plus a feature_scaling.py for preprocessing
• “Helper files”: standalone data-analysis & visualization utilities (null_values_check, pie_chart, etc.)
• “old datasets” & “revised datasets”: raw & processed CSV files
• fig: static images used in README/Reports
• Reports/Misc: documentation, presentations, academic papers
• Config files: requirements.txt, .pre-commit-config.yaml, .gitignore
README for additional insights
• ML stack: Python 3.6+, scikit-learn 1.3.2, XGBoost, GridSearchCV
• Preprocessing steps: missing-value imputation, encoding, scaling, log transforms (feature_scaling.py)
• Feature engineering: ratio-based & binary features
• Model types: Linear Regression, Decision Tree (+ bagging), Random Forest, Gradient Boost, XGBoost
• Deployment: local CLI and Streamlit web app