SPSS command: SPSS FIT
Forecast error statistics for any fitted series
FIT compares an observed series with a fitted or forecast series and summarises the errors.
Free · No account · Files stay on your device · Updated 18 October 2026
What it tells you
Use it to score a model from another tool, a hold-out forecast or a simple benchmark: Crosstabs reports the SPSS FIT error statistics, from mean error and MAPE to RMSE and Durbin-Watson.
- How far off the fit is on average, in the series' own units (MAE, RMSE) and in percent (MAPE).
- Whether the fit is biased: a mean error far from 0 means it is consistently too high or too low.
- Whether the errors are autocorrelated, from the Durbin-Watson statistic.
When to use it
Use it when
- You have forecasts from another source (a spreadsheet model, a vendor, a past report) next to the actual values.
- You want to compare several models on the same hold-out periods with identical error definitions.
- You need a naive benchmark (last value or same month last year) to judge whether a model adds anything.
Use something else when
- You have not built the forecast yet: TSMODEL, ARIMA and exponential smoothing already report their own fit statistics.
- The observed series contains zeros and you need MAPE: percent errors are undefined there, so rely on MAE or RMSE.
- You need a significance test comparing two forecasts; FIT reports descriptive error statistics only.
How to run it in Crosstabs
- 01
Open Goodness of fit
In Statistics mode, choose Goodness of fit (FIT) in the Forecasting group.
- 02
Define the time axis
Both series must live in the same file on the same time axis, so each observed value lines up with its fitted value.
- 03
Choose Observed series and Fitted (predicted) series
Pick the actual values as the Observed series and the model output as the Fitted (predicted) series.
- 04
Set Hypothesis df
Enter Hypothesis df (model parameters) to divide the mean squared error by n minus that number; leave 0 to divide by n.
- 05
Run
Click Run. Only time points where both series have a value are used; the footnote gives the count.
Worked example
Computed by the Crosstabs engine from the synthetic data below when this page was built — not typed in.
Data
48 synthetic monthly values (January 2022 to December 2025) of a brand-tracker metric in percent: a level of 40 rising 0.25 points a month, a fixed seasonal pattern that peaks in December and dips in July, and normal noise (SD 1.1) from a fixed seed, so every build produces the same series. The fitted series is a seasonal naive forecast, the value from the same month a year earlier (CREATE lag 12), so the first 12 months have no fit.
| Statistic | Value |
|---|---|
| Mean Error | 2.953 |
| Mean Percent Error | 6.210 |
| Mean Absolute Error | 3.031 |
| Mean Absolute Percent Error | 6.380 |
| Sum of Squared Errors | 403.290 |
| Mean Squared Error | 11.203 |
| Root Mean Square Error | 3.347 |
| Durbin-Watson | 0.438 |
Error = observed − fitted, over 36 time points where both exist. Mean squared error divides by n.
How to read it
Over the 36 months with a fit, the seasonal naive forecast misses by 3.03 points on average (MAPE 6.38%). The mean error of 2.95 is positive, so it under-forecasts systematically: it ignores the upward trend. A Durbin-Watson of 0.44 shows the errors are also autocorrelated, so a trend-aware model should beat this benchmark.
Output tables explained
- Error Statistics
- Mean Error, Mean Percent Error, Mean Absolute Error, Mean Absolute Percent Error, Sum of Squared Errors, Mean Squared Error, Root Mean Square Error and Durbin-Watson, with error defined as observed minus fitted.
- Summary above the table
- A one-line summary above the table states the mean absolute error and the mean error, the two numbers most readers need.
Assumptions and notes
- Observed and fitted values refer to the same time points; misaligned series give meaningless errors.
- Percent errors need non-zero observed values; when any observed value is zero they are left blank.
- Durbin-Watson uses consecutive available errors, so gaps make it approximate.
How the results are validated
Mean error, mean percent error, MAE, MAPE and Durbin-Watson match a numpy reference to 12 decimals, MSE and RMSE to 10 and SSE to 9, including the MSE with a hypothesis df of 2. Validation evidence
Questions
MAE, RMSE or MAPE: which should I report?
MAE is easiest to explain, RMSE penalises large misses more, and MAPE compares series on different scales but breaks down near zero. Reporting MAE with MAPE covers most audiences.
What does the Hypothesis df do?
It sets how many parameters the model estimated, so the mean squared error divides by n − df instead of n, as in SPSS FIT's DFH keyword. It changes MSE and RMSE only.
How do I build a naive benchmark?
Use Create time series with Lag 1 (last period) or Lag 12 (same month last year), download it, add it to your file, then select it as the Fitted (predicted) series.
Is my data uploaded when I run FIT?
No. Both series are read and compared in your browser; nothing is uploaded to Crosstabs servers.
Related procedures
- Time Series Forecasting (Expert Modeler)TSMODEL
- Exponential SmoothingEXSMOOTH
- ARIMAARIMA
- Create Time SeriesCREATE
- Autoregression (AREG)AREG
- Free SPSS alternative: full coverage list
Run Forecast Goodness of Fit on your own data
Open a .sav, CSV or Excel file. Everything runs in your browser; the file never leaves your device.
SPSS is a trademark of IBM Corp. Crosstabs is not affiliated with or endorsed by IBM.