Examples
Examples
Try the starting example
- JSON Predictions [probability of positive,outcome 0/1]
- [[0.1,0],[0.4,1],[0.6,0],[0.9,1]]
- Equal-width probability bins (count)
- 2
- Calibration bins [lower,upper,count,mean probability,positive rate]
- [[0,0.5,2,0.25,0.5],[0.5,1,2,0.75,0.5]]
- Expected calibration error
- 0.25
- Brier score
- 0.185
Try another scenario
- JSON Predictions [probability of positive,outcome 0/1]
- [[0.1,0],[0.4,0],[0.6,1],[0.9,1]]
- Equal-width probability bins (count)
- 2
- Calibration bins [lower,upper,count,mean probability,positive rate]
- [[0,0.5,2,0.25,0],[0.5,1,2,0.75,1]]
- Expected calibration error
- 0.25
- Brier score
- 0.085
Calculation method
Calculation method
Calibration bins [lower,upper,count,mean probability,positive rate]
Bin by floor(p×bins); p=1 belongs to last bin; omit empty bins.
Expected calibration error
Sum bin_fraction × absolute(mean_probability−positive_rate).
Brier score
Mean squared difference between probability and binary outcome.
Symbols and units
- a
- JSON Predictions [probability of positive,outcome 0/1]
- b
- Equal-width probability bins (count)
Use the units shown. A list keeps the order in which you enter its values. The symbol ^ means “raised to the power of”.
Assumptions & limits
Assumptions & limits
- Binary equal-width calibration audit. ECE depends on chosen bins and can hide within-bin error; descriptive on supplied observations. Supported limits: Nonempty 1–10000 rows; 1–100 bins.
Are these inputs private?
These calculations run in your browser. A shared link contains the values you choose to share. Only questions submitted to the separate AI finder are sent to OpenAI.
Further reading
Background reading; this does not imply endorsement or independent review of SumFizz.