Correlation Heatmap Done Right
Avoid misleading heatmaps.
The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.
Help me build a correlation heatmap that doesn't mislead. Variables and types: [VARIABLES]. Sample size and missing data: [SAMPLE_MISSING]. Purpose: [PURPOSE]. Explain which correlation measure suits each variable type, how missing data is handled, and why a correlation matrix may hide non-linear relationships and outliers. Provide code to compute the matrix, order variables by clustering, mask the upper triangle, annotate values, and use a diverging colour map centred at zero. Include a companion scatterplot matrix for the strongest pairs. Add a caution about causation and multiple comparisons and a checklist for reporting. Suggest confidence intervals for key pairs. Add a worked example with five invented variables and the matrix it might produce, showing how one outlier can create a strong correlation, and a rank correlation comparison. Include a paragraph for reporting the matrix honestly, naming the sample size, the measure and the handling of missing data. Also add a two-sentence template for describing the strongest and weakest relationships in a report without implying cause.
Fill in before use
Optional: type your details and the prompt above updates. Nothing is saved or sent.
Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.