seldonian.utils.stats_utils

Functions

custom_cumprod(x)	Custom implementation of np.cumprod that works with autograd Source: https://github.com/HIPS/autograd/issues/257
default_supervised_initial_solution_fn(...)	A wrapper around model.fit() to use as the initial_solution_fn parameter to the spec.
softmax(x)	Calculate the softmax for a vector of values, x
stddev(v)	Sample standard deviation of the vector v, with Bessel's correction
tinv(p, nu)	Returns the inverse of Student's t CDF using the degrees of freedom in nu for the corresponding probabilities in p.
weighted_sum_gamma(arr[, gamma])	Calculate weighted sum of an array, where weights are gamma**(index of arr).