Basic Modelling Techniques

Linear regression is the simplest and most widely used statistical technique for predictive modelling analysis. It is standard in nearly every field of data science as a go to technique. It is typically applied when explaining the relationship between a dependent variable (target) and one or more explanatory variables(predictors) using a straight line. We would likely use Ordinary Least Squares to model a line of best fit. There are two types of linear regression - Simple and Multiple. These can be used to predict the magnitude of a target variable. Linear regression deals with continuous variables and not Bernoulli variables. The problem of Linear Regression is that these predictions are not sensible for classification problems since the true probability must fall between 0 and 1. When outcomes are binary we are dealing with classification. An example of a classification problem is determining whether a passenger survived or perished as in the case of the Titanic. There are no in-betweens but given independent data variables we can predict the probability of survival for given passenger using logistic regression. Probabilities predict by logistic regression are bounded by 1 and 0.