Quick Start¶

Local Installation¶

Install the ADS library using pip (as shown below) or through OCI Conda Packs (see Installation).

python3 -m pip install "oracle_ads[forecast]"

Installation in a Notebook Session¶

Open Environment Explorer and search for AI Forecasting.
Install the relevant Conda pack.
Activate the Conda environment. For example:

odsc conda install -s forecast_v3
conda activate /home/datascience/forecast_v3

Initialization¶

Initialize your forecast project using the ADS CLI command. This command will create several configuration files that can later be used to run the operators on OCI Data Science Jobs.

ads operator init -t forecast --output my-forecast

Input Data¶

Within the my-forecast folder created above, you’ll find a forecast.yaml file. Update this file with the details of your data and forecast. Below is an example using Prophet’s Yosemite Temperature dataset:

cd my-forecast
vi forecast.yaml

kind: operator
type: forecast
version: v1
spec:
    datetime_column:
        name: ds
    historical_data:
        url: https://raw.githubusercontent.com/facebook/prophet/main/examples/example_yosemite_temps.csv
    horizon: 3
    model: prophet
    target_column: y

There are many more options available in this YAML file.

Running the Forecast¶

Run the forecast locally using the following command:

ads operator run -f forecast.yaml

Viewing Results¶

If the YAML configuration does not specify an output directory, all results will be placed in a new folder called results. The performance summary is provided in the report.html file, and the full forecast is available in the forecast.csv file.

open results/report.html