Hola chicos,

Last time I was preparing myself for a job interview, so that's why I focused mostly on R, but I am back with statistical learning material!

So we have discussed linear regression models for simple regression, then t-statistics which serves to determine whether there is a relationship between variables and after that we went through the measurement of model quality. So all of that was designed for one variable regression and today I would like to discuss multiple linear regression .

First of all, what is the point of having multiple regression? Let’s consider an example, where we want to predict how variables such as lecture attendance, number of the book read, exercises done and IQ are predicting the result which is exam score at e.g. Statistics. Having described by me in previous post tools – simple linear regression, we can try to predict the impact of each of the variables for the score. But as all of us students know, success on the exam consists of diverse approaches and combination of mentioned above factors. So here comes multiple linear regression, which will be given by the equation:

Y = β₀ + β₁X₁ + β₂X₂ + … + β_nX_n + ε, where X_p is a single predictor variable.

How to estimate coefficients?

As in linear regression, we will estimate parameters using the least-squares approach and we minimize the sum of squared residuals. RSS formula below:

Okay, todo for today. In the next post we will ponder about questions that are needed to be answered for multiple regression such as model fit and determination whether there is a relationship between response and predictors. So see you soon!

xoxo,

szarki9