# Discovering time series forecasting

A **time series** is a sequence of data points mapped at a certain time frequency. For example, it can contain the price of a stock collected at regular intervals, the sales volumes of each day in a specific timeframe, or the temperature measured by an **Internet of Things** (**IoT**) sensor during the day.

In the following diagram, you can see a time series graph of the data collected by a temperature sensor in the first 15 days of 2020:

**Forecasting** using time series analysis comprises the use of machine learning models to predict the future values of a specific measure by leveraging the known past data.

BigQuery ML provides a specific machine learning algorithm to forecast numerical values using time series data. The algorithm is designed and developed to do the following:

- Automatically clean and adjust the training data to overcome the data...