Shapiro-Wilk test fails with these data:
[14.587726599333871, 14.845914188997108, 14.491111891829638, 13.576509016063552, 14.564079222003288, 14.177274119338016, 14.44720160004999, 13.589365383738828, 13.375261101861817, 13.890673980094979, 14.291088854951136, 13.645218475691312, 14.501583506358715, 15.170848998952321, 15.95745228080566, 16.360081935716625, 17.4103349063546, 17.1850436261102, 16.848565023259475, 17.888398765992683, 16.586566607302917, 16.718486483399357, 15.390015866469467, 16.300712312358275, 17.017659229149473, 16.416757020607307, 12.329070437414767, 12.29630393917019, 10.685393314266232, 9.660898885542535, 8.285611195973566, 7.965889830278456, 8.320972602691585, 8.229170627290415, 11.75215260141066, 13.786596891598476, 12.617404541667204]
while it finishes in R.
Version: 0.4.4
File: NormalityTest.java
Line 82.
Anonymous
This is because NormalityTest.Shapiro-Wilk requires presorted array as an input. The data you entered in is not sorted. I noticed that this is a lack of documentation issue. As a workaround, sort the array before you feed it into shapiro_wilk_statistic.
In version 0.4.5 onwards, there will be an option to request the function to sort the array first. But I will not release 0.4.5 with this bug as the only fix. I will combine it with several items.