A statistical model that predicts the values of an outcome variable using one or more predictor variables. A predictive model can be a verbal summary of a table (for example, a table showing weight by height can be used to make predictions). However, most commonly the term is used to describe models that make predictions using more complicated techniques, such as:

- Regression.
- Tree-Based Models.
- Neural networks.

A predictive model either performs classification or regression depending upon whether the outcome variable is categorical or numerical. In either case the predictor variables may all be categorical, all numerical or a combination of categorical and numerical. The process of training a model is known as supervised learning because the model aims to make its predictions the same as the known outcome variable. Having trained a model it can then be used to predict the outcome variable for either the same data that was used for training or for a different data set. If a different data set is used then the categorical predictor variables must not consist of more categories than those that were used for training, or else a warning is given and NA is predicted for those cases.

## Comments

0 comments

Article is closed for comments.