Table of Contents
No, logistic regression cannot be performed in Google Sheets. Logistic regression is a type of machine learning algorithm which requires complex calculations not supported by Google Sheets. However, other types of regression can be performed in Google Sheets. For example, linear regression can be easily done in Google Sheets using the SLOPE and INTERCEPT functions.
Logistic regression is a method we can use to fit a regression model when the response variable is binary.
The following step-by-step example shows how to perform logistic regression in Google Sheets.
Step 1: Install the XLMiner Analysis ToolPak
To perform logistic regression in Google Sheets, we need to first install the free XLMiner Analysis Toolpak.
To do so, click Add-ons > Get add-ons:
Next, type XLMiner Analysis ToolPak in the search bar and click the icon that appears:
Lastly, click the blue Install button.
Step 2: Enter the Data
Next, we’ll enter the following data into Google Sheets:
We will fit a logistic regression model that uses points and assists to predict whether a basketball player gets drafted into the NBA (0 =No, 1 = Yes).
Step 3: Perform Logistic Regression
To fit the logistic regression model, click the Extensions tab, then click XL Miner Analysis ToolPak, then click Start:
Once you click OK, the summary of the logistic regression model will be displayed:
The coefficients in the output indicate the average change in log odds of getting drafted.
For example, a one unit increase in points is associated with an average increase of 0.212 in the log odds of getting drafted.
The sign on the coefficients tells us whether there is a positive or negative associated between each predictor variable and the response variable.
For example, since points has a positive sign for the coefficient it means that increasing the value for points increases the chances that a player gets drafted (assuming assists is held constant).
Conversely, since assists has a negative sign for the coefficient it means that increasing the value for assists decreases the chances that a player gets drafted (assuming points is held constant).
The p-values in the output also give us an idea of how effective each predictor variable is at predicting the probability of getting drafted:
- P-value for points: 0.02
- P-value for assists: 0.35
We can see that points seems to be a statistically significant predictor variable since it has a p-value less than .05, but assists does not seem to be statistically significant since it does not have a p-value less than .05.