graph to show correlation between two variables

The closer to 100% the more variation in y is attributed to x, and not some outside variable.

You can think of this as a scale of 0 to 100%, the percentage of variation (or changes) in y that can be explained by x. Use the R-Squared value as a sniff test to determine how well this model predicts y from x. The equation enables you to predict how changes in your x variable (sales) will change your y (profit). In this situation, a very low P-value means that you can have greater trust in the Tableau correlation between sales and profit for a customer in any of our particular segments, and that the results we are seeing did not occur randomlyĥ. If it’s higher than that, the Tableau correlation between the variables isn’t statistically significant. You want a p value that is less than 0.05. When you mouse over the line, you will be given an equation and a p-value. Uncheck “Show Confidence Bands.” But leave “Allow a Trend Line per Color” since we only have 4 segments.Ĥ. Hover over a line and click edit trend lines. At the moment, we just want the Tableau correlations, not the confidence bands (which is why you have so many lines).ģ. Several lines will now appear on your graph. Right click on your scatter plot and click Trend Lines>Show Trend Lines.Ģ. So let’s look at a few basic statistical features.ġ. Once you have a sense of what’s affecting your numbers, you can then talk your conclusions to your colleagues and management. While these can sometimes be confusing to an end user who doesn’t have much experience with stats, it’s very helpful to you as an analyst in really knowing what’s going on. If you want to add more analytical and statistical rigor to your analysis, you can add trend lines and various statistics to the view. Trend Lines and Statistical Features Add Rigor to Tableau Correlation Analysis Mousing over that, we see that it’s a particular Consumer customer that has bought over $117k of products from us and has a profit of $34k. We see, for example, one dot up at the top. We now have each of the customers encoded by their segment.Ĥ. The goal would be to have everyone with both high sales and high profits, which would cluster the dots at the upper right corner of the graph.ģ. We’ll now have a dot for every customer that plots both their sales and their profit. Drag Customer Name out into the quadrant. You can easily swap these axes using the swap icon at the top.Ģ. This will build a quadrant with two axes, with Sales along your x-axis as your independent variable, and Profit on your y-axis as your dependent variable. Drag Sales to Columns and Profits to Rows. Scatter Plots to Find Correlation in Tableauġ. Tableau offers several analytical tools to do this. This is Tableau correlation analysis at work. And with enough data, you could probably start to have a pretty good idea that if a man is 6’0 tall he will weigh within a certain range. For example, as height in men increases, so typically does weight. Correlation analysis in Tableau compares two or more quantitative variables to see if values in one vary systematically with values in another. However, looking at correlation in Tableau by looking between numbers, and how one metric affects another, is an extremely valuable skill in analytics. Up to this point, we’ve mostly looked at how data can be segmented by some dimension or over time.

Graph to show correlation between two variables