Rain Prediction

About The Project

Data source:
https://www.kaggle.com/jsphyg/weather-dataset-rattle-package
The dataset contains 10 years of daily weather observations from many locations across Australia.

This project was built with:

• Scikit-learn
• Matplotlib
• Seaborn
• Pickle

Project Summary

1. Data Understanding
   • Descriptions of variables
2. Exploratory Data Analysis
   • Exploring the categorical and numerial variables
   • Feature engineering of Date variable
   • Outlier detection using boxplots
   • Checking the distribution of numerical variables using histograms
   • Checking the distribution of target variable (class distribution)
   • Correlation analysis
   • Checking for duplicates
3. Data Pre-processing
   • Handling missing values
   • Removing outliers
   • Categorical data encoding
   • Feature scaling
   • Feature selection
4. Training Baseline Models
   • Created a function to evaluate performance of multiple models using multiple metrics through Cross Validation

   

5. Shortlisting the Best Models
   • Selected the top 3 models
6. Hyperparameter Tuning
   • Determined the best parameters of the models using Randomized Search Cross Validation
7. Building Ensemble Models
   • Built 3 ensemble models using Stacking Classifier
8. Model Evaluation
   • Evaluated the performance of the 3 inital shortlisted models and the 3 ensemble models
   • Plotted learning curves to compare the performance of the models on training and testing data
   • Determined the best model

Collaborators

• Andy Chow Sai Kit
• Wong Yew Lee
• Li Chen Zhen