Machine learning enables you to use data from the past to predict future outcomes. AutoML Tables provides predictions for either regression or classification problems. Understanding what kind of problem you are solving (and what you want to predict for) helps you create an effective dataset.
A regression model predicts a numeric value. Some examples of regression problems include how much will a house sell for or how much will a given consumer spend on a website in the next month.
A regression model has a target column that is Numeric.
A classification model predicts a category from a discrete, fixed number of possible categories. Some examples of classification problems include whether an email is spam or not, or what type of classes a student might be interested in. If your target column's value can be one of three or more choices, then you have a multiclass classification problem.
A classification model has a target column that is Categorical.