3.9.4 Prediction Intervals

3.9.4.1 Using Forecasting Models with Prediction Intervals

MySQL 8.4.0 introduces prediction intervals for forecasting models, which generate upper and lower bounds on predictions for forecasting based on level of confidence. For example, for a prediction interval of 0.95 with a lower bound of 25 units and an upper bound of 65 units, you are 95% confident that product ABC will sell between 25 and 65 units on a randomly selected day.

When using prediction intervals, the prediction_interval option is included for the ML_PREDICT_TABLE routine, which specifies a level of confidence. Predictions provide three outputs corresponding to each endogenous variable: the forecasted value, a lower bound, and an upper bound. The prediction_interval option can be provided as a JSON_OBJECT.

For the prediction_interval option:

• The default value is 0.95.

• The data type for this value must be FLOAT.

• The value must be greater than 0 and less than 1.0.