The Kolmogorov-Smirnov Test for Goodness of Fit

By F. Massey. 1951

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Evaluation & Benchmarks

Citations in the corpus

On The Impact of Machine Learning Randomness on Group Fairness (2023, methodology): “We also perform the Kolmogorov–Smirnov (KS) test [33] to match sampling across multiple runs (at epoch 300) and sampling across epochs (for a single training run with seed 0).”
The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks (2018, methodology): “When we run the Kolmogorov–Smirnov goodness-of-fit test [32] on 106 samples, we fail to reject the null hypothesis (p > 0.”
On Network Design Spaces for Visual Recognition (2019, methodology): “We can make this comparison quantitative using the (two sample) Kolmogorov-Smirnov (KS) test [24], a nonparametric statistical test for the null hypothesis that two samples were drawn from the same distribution.”
Thresholded ConvNet ensembles: neural networks for technical forecasting (2018, methodology): “For our purposes, we define a technical pattern to be informative if its presence significantly alters the distribution of next-day returns in a Kolmogorov–Smirnov two-sample (K–S) test [16], comparing the unconditional distribution of all next-day returns to the distribution conditioned on having j”
The Atacama Cosmology Telescope: A Measurement of the DR6 CMB Lensing Power Spectrum and Its Implications for Structure Growth (2023, methodology): “The Kolmogorov–Smirnov statistic (Massey 1951) shows that both distributions are well described by Gaussians, with PTEs of 0.6 and 0.9 respectively.”
The Synthetic Data Vault (2016, methodology): “In order to test for this fit, we use the Kolmogorov-Smirnov test [3], which returns a p-value representing the likelihood that the data matches a particular distribution.”
Learning grounded finite-state representations from unstructured demonstrations (2015, methodology): “The difference in the length distributions is tested using the Kolmogorov-Smirnov test [Massey, 1951] as follows.”
Small vulnerable sets determine large network cascades in power grids (2017, unknown): “The Kolmogorov-Smirnov test (47) cannot reject the hypothesis that the size of cascades triggered by nt = 3 line failures has the same underlying distribution as of those triggered by 4 line failures (p-value = 0.”
History and trends in solar irradiance and PV power forecasting: A preliminary assessment and review using text mining (2018, methodology): “As shown in Table 2, KSI essentially evaluates the area between two empirical CDFs.”
Maximum entropy-Gumbel-Hougaard copula method for simulation of monthly streamflow in Xiangxi river, China (2015, background): “S test quantifies the largest vertical difference between the specified and empirical distributions (Massey 1951; Razali and Wah 2011).”