Food Delivery Speed Classification

Developing an end-to-end multiclass classification pipeline to predict food delivery speeds (**Slow, Average, Fast**) based on environmental, route, and order-level features. Using a dataset of 1,000 order records, Exploratory Data Analysis (EDA) was performed, addressing data quality issues through median imputation for missing values, categorical feature encoding, and feature standardization via `StandardScaler`. To mitigate significant class imbalance across target categories, **SMOTE** was applied to ensure reliable model training across all speed tiers. Multiple algorithms—including Logistic Regression, Neural Networks, and Decision Trees—were trained and benchmarked, with performance evaluated using Confusion Matrices, Precision, Recall, and ROC-AUC metrics. **Logistic Regression emerged as the top performer**, achieving **78% accuracy** and a **0.92 macro ROC-AUC score**, demonstrating that regularized linear models can effectively capture key predictors like distance and preparation time without the overhead of complex architectures.