To build a short demo for AWS Forecast, you’ll want to focus on the key steps involved in setting up, training, and generating predictions. Here's a simple guide to create a short demo:

1. Define the Scope of the Demo (5-10 minutes)

• Objective: Show how to upload historical data, train a model, and generate predictions using AWS Forecast.
• Example Use Case: Predict demand for a product based on historical sales data.

2. Prepare Your Data (1 minute)

• Use a simple time-series dataset (e.g., historical sales, demand data, or temperature).
• Format the dataset as CSV with columns such as timestamp, item_id, and demand.
• Make sure the dataset is in a format acceptable by AWS Forecast (e.g., Amazon S3 CSV file).

3. Create a Dataset Group and Dataset (1-2 minutes)

• Go to the AWS Management Console → AWS Forecast.
• Create a Dataset Group and choose a domain (e.g., Retail, Custom).
• Create a Dataset and specify its schema (timestamp, item_id, demand).
• Upload your dataset from Amazon S3.

4. Import the Data into AWS Forecast (1 minute)

• Use the Data Import wizard to import your data from S3 into the created Dataset Group.
• Show the progress of data import in the console.

5. Train a Forecast Model (2 minutes)

• Choose Predictor → Create Predictor.
• Select AutoML for AWS to automatically select the best model.
• Configure the training settings (forecast horizon, frequency).
• Start training the model and explain that AWS Forecast will analyze the dataset and pick the optimal algorithm.

6. Generate Forecasts (1 minute)

• Once the model is trained, create a forecast by selecting the trained model.
• Generate predictions for future time periods (e.g., demand for the next 30 days).
• View the forecast output in the AWS console or export it to S3.

7. Visualize or Access the Forecast Data (1-2 minutes)

• Show how to access forecast data via the AWS console or API.
• Optionally, visualize the forecast using QuickSight or a simple Python script to plot the results.

8. Wrap-up (1 minute)

• Summarize how AWS Forecast can be used to easily generate forecasts using historical data.
• Mention other advanced features (AutoML, accuracy metrics, or anomaly detection) for future exploration.

---

Demo Outline:

1. Introduction: Objective (0:30 sec)
2. Prepare Data: Show formatted dataset (1 min)
3. Create Dataset Group and Import Data: (1-2 mins)
4. Train Model: Using AutoML (2 mins)
5. Generate Forecast: Export and visualize (1-2 mins)
6. Wrap-up: Benefits of AWS Forecast (1 min)

This short demo ensures you cover the basics of AWS Forecast while keeping it under 10 minutes.

References

1. https://docs.aws.amazon.com/forecast/latest/dg/getting-started.html