How to use correlation matrix

Correlation matrix and scatter plot matrix

When reviewing data collected in marketing or R&D, the first step in data analysis is to understand the characteristics of the data before applying regression models. Understanding the relationships among variables in the data is particularly important. The correlation coefficient is a representative statistic that expresses this relationship.

This function outputs scatter plots and correlation coefficients for all combinations of variables in the data at once and visualizes them as a "scatter plot matrix" and a "correlation matrix. This is a very useful tool for quickly understanding the characteristics of the data.

Ex. Correlation check of sales data

As an example, let's assume sales log data from a hypothetical retail store.

Enter the csv file, select all the variable columns for which you want to compare correlations, and press the Output button to create a scatter and correlation matrix chart in the Output tab. Note that if you select columns that contain non-numeric data, you will not be able to create them.

The top part of the output tab displays the scatterplot matrix and the bottom part the correlation matrix.

This data includes customer age and sales amount, as well as the amount of transactions and time spent online and in the physical store. The characteristics between these variables can be checked at once.

For example, the correlation matrix shows that there is a strong positive correlation between online transaction value and time spent.

For those correlations that are close to zero, it is important to check the scatter plots to see how they are distributed. For example, a scatter plot of online time spent and in-store time spent shows that they are strongly inversely correlated. The hypothesis is that most customers use only one of the two.

In reality, there are many more variables to compare, such as the details of customer data and product categories purchased. By creating a matrix chart with this function, you can quickly see the big picture of your data, formulate a hypothesis or further analysis strategy, and proceed smoothly with your analysis.