Evaluating the time series forecasting model
In this section, we'll evaluate the performance of the machine learning model that we trained in the previous one.
The evaluation stage of a time series model can be performed by using the ML.EVALUATE BigQuery ML function.
Let's execute the following query to extract all the evaluation parameters that characterize the ARIMA model:
SELECT * FROM ML.EVALUATE(MODEL `08_sales_forecasting.liquor_forecasting`);
The results of the query are visualized in the following screenshot:
Figure 8.13 – The records extracted from the evaluation of the time series forecasting model
Each row defines each non-seasonal ARIMA model classified as an ARIMA(p,d,q) model. For each row, we can notice the following:
- The field non_seasonal_p represents the parameter p of the ARIMA model. The value of the row is the number of autoregressive terms used for the prediction. It indicates the number of observations...
