Recipe Rating Prediction
A project with data exploration, feature engineering, recommendation models based on Random Forest and Logistic Regression to predict recipe ratings, addressing challenges like data imbalance and sparse interactions.
This project predicted recipe ratings using Food.com data, leveraging a mix of feature-based (e.g., regression, Random Forest) and interaction-based models (e.g., collaborative filtering). Preprocessing included outlier removal, data balancing via SMOTE, and Word2Vec embeddings for textual features. Key findings showed user and recipe averages as critical predictors. Logistic regression and Random Forest models excelled, while interaction-based approaches were hindered by data sparsity. Future work suggests incorporating sentiment analysis and neural collaborative filtering for enhanced performance.
The detailed report is available here. The code is available on GitHub.