Machine Learning with Swift

Understanding and overcoming the limitations of linear regression

Before building the predictive model, you should always perform an exploratory analysis. It will help you to select the right model by identifying the relationship and impact of features and samples. Linear regression has a whole bunch of preconditions and hidden assumptions. To get accurate results, you need to be sure that all those conditions are met and all assumptions are true:

• Linear regression assumes all features to be numerical variables. If you have categorical features, you cannot use linear regression. You need to be careful here, because often categorical variables are represented by numbers; for example, country code or E numbers of food additives (found on all ...