CMPE-255-Clustering-Assignment

This repository contains Colab notebooks demonstrating the capabilities of TimeGPT, Tabular, and RelBench for various machine learning and time series forecasting tasks.

A single video tutorial explains the complete assignment, providing insights into each notebook's functionality and the overall workflow.

Requirements

Before using this repository, ensure you have the following installed:

• Python 3.8+
• Google Colab (or Jupyter Notebook if running locally)
• GitHub CLI (optional, for easy artifact uploads)

Video Tutorial: Youtube Video Link

Notebooks Overview

TimeGPT Notebooks

1. Multivariate Time Series Forecasting

   • Learn to set up multivariate time series data and use TimeGPT for predictions.

2. Long-Horizon Forecasting

   • Explore extended period predictions with TimeGPT.

3. Fine-Tuning with Custom Data

   • Fine-tune TimeGPT with your dataset for better performance.

4. Anomaly Detection

   • Detect anomalies in time series data using TimeGPT.

5. Energy Forecasting

   • Predict energy demand using TimeGPT.

6. Bitcoin Price Prediction

   • Forecast Bitcoin prices with TimeGPT using real-world datasets.

Tabular Notebooks

1. Synthetic Data Generation

   • Use Tabular to create synthetic datasets with real-world distribution.

2. Zero-Shot Inference

   • Perform zero-shot inference with pre-trained Tabular models.

RelBench Notebook

1. GNN for Tabular Prediction
   • Utilize RelBench to train and evaluate a Graph Neural Network (GNN) for tabular prediction tasks.