Add a distribution protocol about statkit HOT 3 CLOSED

The Distribution protocol should be very general, encapsulating commonalities from univariate, multivariate, continuous, and discrete distributions. Additional protocols can then refine the Distribution protocol to add specialized functionality. Refining protocols could be MultivariateDistribution, DiscreteDistribution, and ContinuousDistribution.

from statkit.

Here's a suggested initial split of functionality between the different protocols.

Distribution

• Cumulative Distribution Function
• Reliability Function
• Percent Point Function
• Inverse Reliability Function
• Mean
• Variance
• Standard Deviation
• Skew
• Kurtosis
• Non-central and Central Moment

Multivariate Distribution

• Covariance

Continuous Distributions

• Probability Density Function

Discrete Distributions

• Probability Mass Function

from statkit.

We'll leave out the multivariate refinement in this story. As it looks, it's a purely additive feature and will not be a problem to incorporate at a later time.

from statkit.