The correlation tool allows you to correlate a primary numerical data series against one or many secondary numerical series. The correlation analyzer will analyze each of your secondary series separately against your primary series, displaying the results of each correlation in a table view and in a graph.
Select a Primary Series¶
To create a new correlation, you'll need to first select the primary series that you care about. This series will be compared with all other series independently. The series you select must have all numerical data, and the number of rows in this series must match the number of rows in all secondary series.
Select a Secondary Series¶
Once you select a primary series, you can then select multiple secondary series. Each of these will be independently correlated with the primary series. If you select a series in a dataset that is different from the dataset in which the primary series exists, you'll be asked to join the first and second data sources together. This means that you need to select the column in the first dataset whose values will also exist in the second dataset.
For example, if you're trying to correlate the stock returns of two stocks in two different data sets, you'll need to select the column in both datasets that represents the date for each stock return.
When you click the "Correlate" button, the results page will show you a graph section and a table section. The graph will show a scatterplot of the correlation between the primary series and the first of the secondary series. You can select a different secondary series by clicking on the dropdown and select a new series.
The table below the graph will show you a list of all of the correlation values and associated p-values. Strong correlations are closer to -1 or +1, with weak correlations closer to 0. P-values closer to 0 indicate stronger statistical significance.
Adding to Dashboards¶
Like other analyses and visualizations, you can save the results of your correlation analysis and add them to your dashboards using the buttons at the top of the page.