How to use custom holidays for time-series forecasting in BigQuery ML

Time-series forecasting is crucial in various industries. Holidays impact time-series data, making it important to account for them in forecasting models. BigQuery ML now offers custom holiday modeling capabilities in ARIMA_PLUS and ARIMA_PLUS_XREG models. These capabilities allow users to access built-in holiday data, customize holiday parameters, and explain the contribution of individual holidays to forecasting results. By creating a custom holiday for an event like Google I/O, users can significantly improve the accuracy of their forecasts. Custom holiday modeling allows users to incorporate company-specific holidays, enhancing the explainability and accuracy of forecasting models. BigQuery ML provides a public dataset and the ML.HOLIDAY_INFO table value function to facilitate understanding of the holidays used in forecasting models. Custom holiday modeling in forecasting models is now available for preview in BigQuery ML. It offers benefits such as ease of configuration using GoogleSQL, enhanced transparency, and improved explainability of time series forecasting.