https://github.com/ranaessam03/climate-prediction

sA machine learning project to predict rainfall using Decision Tree, k-Nearest Neighbors (kNN), and Naïve Bayes, with preprocessing, model evaluation, and custom kNN implementation.
https://github.com/ranaessam03/climate-prediction

decision-trees knn-classification machine-learning naive-bayes-classifier numpy sklearn

Last synced: 8 days ago
JSON representation

sA machine learning project to predict rainfall using Decision Tree, k-Nearest Neighbors (kNN), and Naïve Bayes, with preprocessing, model evaluation, and custom kNN implementation.

• Host: GitHub
• URL: https://github.com/ranaessam03/climate-prediction
• Owner: RanaEssam03
• Created: 2024-11-30T05:20:56.000Z (27 days ago)
• Default Branch: main
• Last Pushed: 2024-12-15T10:19:35.000Z (12 days ago)
• Last Synced: 2024-12-15T11:24:15.089Z (12 days ago)
• Topics: decision-trees, knn-classification, machine-learning, naive-bayes-classifier, numpy, sklearn
• Language: Jupyter Notebook
• Homepage:
• Size: 1.62 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md

Awesome Lists containing this project

README

        
# Climate Prediction

### Overview

This project is part of the Machine Learning course at Cairo University, focused on implementing and evaluating Decision Tree, k-Nearest Neighbors (kNN), and Naïve Bayes algorithms. The task involves predicting whether it will rain using a modified weather forecast dataset obtained from Kaggle. The dataset contains six features and 2,500 observations.

### Problem Statement

The objective is to:

1. Preprocess the data for machine learning.

2. Implement and evaluate machine learning models using scikit-learn.

3. Build the kNN algorithm from scratch and compare its performance to the pre-built version.

4. Interpret and evaluate the models with respect to their decision-making processes and performance metrics.

### Dataset

The dataset consists of the following features:

- **Temperature**

- **Humidity**

- **Wind Speed**

- **Cloud Cover**

- **Pressure**

- **Rain** (target variable: 1 if it rains, 0 otherwise)

### Tasks

#### Task 1: Preprocessing

1. **Missing Data Handling**

   - Identify missing data in the dataset.

   - Handle missing data using two techniques:

     - Dropping missing values.

     - Replacing missing values with the feature’s mean.

2. **Feature Scaling**

   - Check if the data is on the same scale.

   - Apply standard scaling if necessary.

3. **Data Splitting**

   - Split the dataset into training and testing sets (80/20 split).

#### Task 2: Model Implementation

1. Implement Decision Tree, k-Nearest Neighbors (kNN), and Naïve Bayes using scikit-learn.

2. Evaluate models using accuracy, precision, and recall metrics.

3. Implement the k-Nearest Neighbors (kNN) algorithm from scratch.

4. Compare the custom kNN implementation with scikit-learn’s kNN using the evaluation metrics.

#### Task 3: Interpretation and Evaluation

1. **Effect of Data Handling**

   - Evaluate the performance of the models under different missing data handling techniques.

2. **Decision Tree Explanation**

   - Visualize the decision tree.

   - Explain the criteria and logic used at each node for predictions.

3. **Performance Metrics Report**

   - Compare the custom kNN implementation with scikit-learn’s kNN using at least five different values of `k`.

   - Provide detailed reports on the accuracy, precision, and recall of all models.

### Dependencies

This project uses the following Python libraries:

- `numpy`

- `pandas`

- `scikit-learn`

- `matplotlib`

### Code Highlights

1. **Preprocessing**

   - Handling missing data by both dropping and replacing with mean.

   - Standard scaling of numeric features.

2. **Model Implementations**

   - Custom kNN implementation using Euclidean distance.

   - Comparison of custom kNN with scikit-learn’s implementation.

   - Decision Tree and Naïve Bayes implementation using scikit-learn.

3. **Visualization**

   - Decision tree visualization for interpretation.

### How to Run the Code

1. Install required libraries: `pip install numpy pandas scikit-learn matplotlib`

2. Place the dataset file `weather_forecast_data.csv` in the same directory as the code.

3. Run the Python script to preprocess data, train models, and evaluate performance.

### Results

- The project evaluates the impact of different missing data handling techniques.

- Performance of models is compared using multiple metrics (accuracy, precision, recall).

- Insights are drawn from the custom kNN implementation and decision tree visualization.

### Future Work

- Extend the dataset with additional features for better predictions.

- Explore advanced hyperparameter tuning techniques for improved model performance.

- Implement additional algorithms to enhance comparative analysis.