This blog post on multiple linear regression is an extension of my blog posts Understanding AI Algorithms and Simple Linear Regression. They are all part of my series of algorithms that can be used in marketing with AI.
Linear regression models containing more than one independent variable are called multiple linear regression models.
Linear regression models containing more than one independent variable are called multiple linear regression models.
For example, your salary may be dependent on both the type of university degree you have and the number of years you have worked, not just the years alone.
Multiple linear regression is useful when seeking to discover the relationship between a set of independent variables and a continuous dependent variable, which are variables that can take any numeric value (like salary, revenue, etc.
Just like simple linear regression, this approach assumes that there is a linear relationship between the variables.
However, instead of having a linear relationship between the dependent variable and one independent variable, this model allows relationships with multiple independent variables.
Let’s consider the mountain climber again. We know that there is a linear relationship between the time a person has spent walking and their elevation.
We can also assume that a person’s height is part of that relationship because taller people go faster with a longer stride. Therefore, the dependent variable – elevation – is dependent on both the amount of time spent hiking and the length of the person’s stride.
With three variables in the model, it uses a broader collection of information, and if graphed, it becomes three dimensional.
Just like simple linear regression, this model allows you to interpret and explain how changes in the independent variables will affect the dependent variable in an easily accessible manner.
This could be useful when wondering how different customer attributes are likely to affect how much they are spending or to predict how much a product will sell based on how much is spent on advertising and what channel is used to reach out to customers.
Multiple linear regression is like simple linear regression in that they are both easy to understand. In the real world, this is useful in supporting business decisions or explaining the effects of a particular course of action. It also allows for quick and easy predictions based on data already collected.
However, this simplicity can create some difficulties.
Data scientists often change the way variables are presented by using equations to alter their value. This helps fit variables into the model, but too much manipulation of the data can weaken the model’s viability and complicate our interpretation of it.
Although a data scientist can manipulate variables to achieve a linear relationship, that may not be enough to explain the data properly.
Many real life problems are more complex than a linear relationship can relay, which means the model may be missing important information.
If you want to read all the related articles on the topic of AI algorithms, here is the list of all blog posts in this article series:
