By Team Edtoolkits · Charts
Calculate pairwise Pearson correlations with sample counts.
Use the guide alongside the working tool.
Open Correlation Matrix Maker →A small dataset to try
Record,Ad spend,Visits,Orders
1,10,100,20
2,20,200,40
3,30,300,60
4,40,400,80These deliberately constructed columns are exact positive multiples, so each pair has r = 1. Real business data rarely behaves this cleanly, and the example does not establish that ads caused visits or orders.
When to choose this chart
A correlation matrix compares pairwise linear relationships between numeric columns. Each coefficient is Pearson r; it is not a causal effect or a guarantee of a useful prediction.
Prepare and map your fields
Use numeric measurements on matching observations. The tool uses pairwise complete rows, so different cells can have different sample counts when values are missing. Constant columns have undefined correlations.
Select comparable observation columns and inspect missing data.
Read each coefficient with its sample count, not only its colour.
Follow up important pairs with a scatter plot to inspect nonlinear patterns and influential points.
Read the result correctly
Values near +1 or −1 indicate strong linear association; a value near zero can still hide a nonlinear relationship. N/A should not be read as zero.
A mistake to avoid
Small samples, outliers and shared time trends can make correlations unstable or misleading. Different pairwise sample sizes also make cells less directly comparable.
Export and keep your data private
Review the title, units, labels and any omitted-row warnings at the final export size. PNG is a raster image, SVG retains vector chart elements, and PDF embeds a raster image. Save the original table if you may need to revise the mapping.
Data stays in browser memory; no upload, external chart service or AI model. CSV, TSV, semicolon/pipe text, JSON and JSON Lines: 2 MB, 10,000 rows, 24 columns. Charts allow up to eight series; categorical charts 60 groups, lines/scatter/candlesticks 500 rows, pie/donut/waffle 12 categories, radar 3–12 dimensions. Blank/invalid mapped values are omitted and counted. Numeric values use a decimal point; remove currency symbols and grouping separators. Percent strings are treated as displayed numeric values (25% becomes 25). Dates use ISO format. Recommendations check data structure, not meaning or causation. PNG up to 16 MP; SVG stays vector; PDF embeds a high-resolution chart image. Review mappings, ordering, aggregation and labels before sharing.
Ready to apply this workflow?
Try Correlation Matrix Maker →Explore a related workflow
- Scatter Plot Maker: Examples & Data Guide
Explore the relationship between two numeric columns.
- Chart Studio: Examples & Data Guide
Find suitable charts for your data and create a downloadable chart collection.
- Column Chart Maker: Examples & Data Guide
Compare categories with grouped columns and a clear zero baseline.
- Bar Chart Maker: Examples & Data Guide
Compare ranked categories with horizontal bars and readable labels.
Keep your original files and review the result. Found unclear instructions? Contact Team Edtoolkits.