Introduction¶

Welcome to our workshop! In this workshop we'll be using the open-sourced IBM Granite AI foundation models for a number of use cases that demonstrates the value of timeseries forecasting and generative AI.

By the end of this workshop, you will be able to:

Predict future trends using time series forecasting
Train a Preprocessor and evaluating the model and evaluate the performance of the model on a real-world dataset containing energy demand data
Move beyond zero-shot prediction to few-shot fine-tuning and prediction
Combine zero-shot inference and fine-tuning with exogenous inputs
Use the watsonx SDK to perform inference calls against a model hosted remotely
Employ few-shot fine tuning, evaluation, and visualization of retail sales data forecasting

The introductory page of the workshop is broken down into the following sections:

Agenda¶


Lab 0: Pre-work	Pre-work for the workshop
Lab 1: Energy Demand Forecasting - Basic Inference	Getting started with TTMs.
Lab 2: Energy Demand Forecasting - Preprocessing and Performance Evaluation	Learn about preprocessing and performance evaluation.
Lab 3: Energy Demand Forecasting - Few-shot Fine-tuning	Learn about few-shot Fine-tuning.
Lab 4: Bike Sharing Forecasting - Zero-shot, Fine-tuning, and Performance Evaluation	Learn about zero-shot forecasting as well as fine-tuning using a Kaggle bike-sharing dataset.
Lab 5: Getting Started with Watson X AI SDK	Use the watsonx SDK to perform inference calls against a model hosted remotely on watsonx.
Lab 6: Retail Forecasting using M5 Sales Data - Few-shot, Fine-tuning, Evaluation, and Visualization	Explore time series forecasting using the IBM Granite Time Series model to predict retail sales.

The technology used in the workshop is as follows:

The notebooks used in this workshop are versions of notebooks from the IBM Granite Community modified for the workshop needs