Write a class that does One-Way ANOVA. You decide what methods it should offer and what variables to maintain. Among the functionalities would be an overall F test, t-tests for user-specified contrasts. We don't have functions for the cdf of the F and t distributions, so you will calculate a p-value from the permutation null distribution obtained by shuffling the group memberships against the y values.. Test your class by instantiating it on artificial data that you generate with random numbers.