Summary

In this chapter, we've built our time series forecasting machine learning model. After the introduction of the business scenario, we discovered what time series forecasting is, and in particular, the ARIMA algorithm that is used to predict values from historical data points.

Before diving into the development of the BigQuery ML model, we applied some analyses on the data collected by the state of Iowa related to liquor sales in the shops of the territory. For this purpose, we introduced the use of the reporting tool Data Studio, which can be easily accessed by the BigQuery UI and be leveraged to draw a time series chart.

We then created our training table, which includes the time series of historical data, and trained our BigQuery ML model on it. Then, we evaluated the time series forecasting model by leveraging the BigQuery ML SQL syntax.

In the last step, we forecasted the quantity of liquor sold in Iowa with a horizon of 30 days and drew the results in a Data...