Kaggle Dataset Fetcher and Predictor

This project provides tools to search for datasets on Kaggle, download and preprocess them, and perform predictions using a Linear Regression model. It includes interactive text-based user interfaces built with curses.

Features

• Search for datasets on Kaggle interactively.
• Download datasets and automatically extract files.
• Load datasets into a pandas DataFrame and preprocess them.
• Train a Linear Regression model and evaluate it using RMSE and MAE.
• Visualize results with scatter plots.

---

Prerequisites

1. Python: Python 3.7 or higher.
2. Install Required Libraries:

pip install pandas numpy matplotlib scikit-learn kaggle

Set Up Kaggle API:

• Go to Kaggle Account.
• Download the kaggle.json API token.
• Place it in ~/.kaggle/ (Linux/Mac) or %USERPROFILE%.kaggle\ (Windows).

File Structure

.
├── kaggle_connect.py  # Handles dataset search and download via Kaggle API.
├── prediction.py      # Performs data preprocessing, model training, and visualization.
└── README.md          # Documentation for the project.

Usage

Step 1: Search and Download a Dataset

Run the following command to search, download a kaggle dataset and prediction script:

python prediction.py

Or

python3 prediction.py

Follow the interactive prompts:

1. Enter a search term for datasets (e.g., Boston Housing Dataset).
2. Select a dataset from the list.
3. Specify a folder to store the downloaded files.

Step 2: Train and Test a Linear Regression Model

The script:

4. Displays descriptive statistics of the data.
5. Splits the data into training and testing sets.
6. Trains a Linear Regression model and evaluates its performance.
7. Displays a scatter plot comparing actual and predicted values.

Example Output

Terminal Interface

Dataset Selection

Model Metrics

Scatter Plot

Contributions

Contributions are welcome! Feel free to submit issues or pull requests to enhance the functionality.