The quality of the description of the fluxes by the model components indicates a goodness-of-fit assessment.Goodness of fit is a statistical measure of how well a model's predictions match observed data. In physics, it is used to judge whether a theoretical curve, a simulation, or a combination of model components adequately explains experimental measurements. A high goodness of fit means the model captures the essential features of the data, while a low value suggests missing physics, systematic errors, or an inappropriate model.
The concept emerged in the late 19th century with Karl Pearson's chi-squared test (1900), which formalized the comparison of observed frequencies to expected ones. It became a standard tool in physics data analysis in the 20th century.
The most common quantitative measure is the chi-squared statistic, which sums the squared differences between observed and predicted values, each weighted by the uncertainty. A reduced chi-squared near 1 indicates a good fit; much larger values signal underfitting or underestimated errors, and much smaller values can indicate overfitting or overestimated errors. Other metrics like the coefficient of determination (R²) or likelihood-based criteria are also used depending on the context.
Used in particle physics to validate the Standard Model against collider data, in astrophysics to fit galaxy rotation curves, and in any experiment where a model is compared to measurements.
In particle physics, a 'good' fit often requires a reduced chi-squared between 0.8 and 1.2, but a value exactly 1.0 is so rare that it sometimes raises suspicion of data fabrication.
