Skip to content
ENExplore Calculators

Data Drift (PSI) Calculator

Work out data drift (psi) instantly with clear inputs, formula shown and shareable results.

About Data Drift (PSI) Calculator

The Population Stability Index sums (actual - expected) x ln(actual / expected) over the bins of a feature, which is a symmetrised relative-entropy measure. Each bin's contribution is positive whether the share rose or fell, so the total grows with any redistribution. The industry convention is that below 0.1 is stable, 0.1 to 0.25 warrants monitoring and above 0.25 signals a material shift.

Formula

PSI

PSI = sum over bins of (actual% - expected%) x ln(actual% / expected%)

Worked example

With the default values below, Data Drift (PSI) Calculator returns PSI contribution from this bin of 0.05640. Change any input above and every figure here is recalculated in your browser.

Inputs used

Expected share in this binaccepts 0.01–100 %
20 %
Actual share in this binaccepts 0.01–100 %
32 %
Number of bins in the featureaccepts 2–100
10

Results

PSI contribution from this bin
0.05640
Estimated total PSI if shifts were similar
0.2820
Log ratio actual / expected
0.47000
PSI band
Major shift (PSI >= 0.25) - investigate or retrain

Step by step

  1. 1Values used
  2. 2PSI
  3. 3PSI contribution from this bin= 0.05640
  4. 4Estimated total PSI if shifts were similar= 0.2820
  5. 5Log ratio actual / expected= 0.47000
  6. 6PSI band= Major shift (PSI >= 0.25) - investigate or retrain

Assumptions and limits

  • The result is derived only from the 3 inputs listed above. Anything not among them is not part of the calculation.
  • In particular, this calculator does not take inflation, tax, fees or charges or compounding frequency into account.
  • Figures are rounded for display to 5 decimal places, so totals built from several rounded parts can differ slightly from the unrounded arithmetic.

How these calculators are built and tested is described on the Methodology page. If a figure looks wrong, please report it.

Frequently Asked Questions

How should bins be chosen?

Use deciles of the baseline distribution so each expected share is 10 percent. Equal-width bins on a skewed feature put almost everything in one bin and hide real drift.

What if a bin has zero actual observations?

The logarithm is undefined, so implementations substitute a small floor such as 0.0001. Report that you have done so, because the PSI value becomes sensitive to that choice.

Related Calculators

Data Drift (PSI) Calculator | Calcrivo