ManishaLagisetty/Natural-Disaster-Prediction-Using-Machine-Learning — reverse-engineered prompt

Build me a complete Python machine learning project that predicts the type of natural disaster from historical disaster data.

Use the included CSV dataset, clean it, handle missing values, encode categories, pick useful features, and save a preprocessed dataset. I want a clear notebook and a runnable Python script that walk through the whole flow in plain English, including simple charts for exploring the data before and after cleaning.

Train and compare several models, including Random Forest, SVM, K nearest neighbors, and Naive Bayes. Show accuracy, precision, recall, and F1 score, then try balancing the data, voting ensembles, and basic tuning to see if the results improve. Save the best model with joblib and include a separate test script or notebook that loads the saved model and runs predictions on a new unseen CSV.

Please keep the code easy to follow, add comments, and make the outputs understandable for someone reviewing this as a college machine learning project.