A linear regression is a process that models the relationship between an independent and dependent set of values in order to find a trend based on historical data (when applied in machine learning).

For example, if you wanted to predict a baseball player's salary based on performance, you can use independent variables, such as batting average, homeruns, etc. and then have the machine learning algorithm be trained on the other players statistics and respective salares in order to predict a player's salary (the dependent variable).