New spire.stats package (take 2) #554
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
In the present structure, random.Dist is a way of producing arbitrary values from a Generator. However a probability distribution should be a way of assign a probability to each measurable subset of the possible outcomes of a random experiment [1], i.e. you should be able to ask it for pdf, cdf, mean etc. The mechanism for production of values according to a probability distribution function should in reality be a random variable [2]. With the proposed separation it should be possible to characterize a probability distribution and still be able to modify the production of values from the Generator (the role of random.Dist would basically go to stats.RandomVariable). stats.Distribution and stats.RandomVariable would be immutable. [1] http://en.wikipedia.org/wiki/Probability_distribution [2] http://en.wikipedia.org/wiki/Random_variable
Extract Anderson-Darling test from random.GaussianTest to be able to use it in the library, with any continuous distribution.
Create location-scale family to help implementation of these families, e.g. uniform and normal distributions. In a location-scale family a general distribution can be written as a function of the standard distribution (zero location and unit scale) and its own location and scale.
Enrich Generator to return the next random value based on a specified type. That way one can ask for a next[Double] or next[Int] instead of nextDouble or nextInt. However there's a name clash, therefore the method name used is next0[A]. Should this method be included in the Generator class itself?
Discrete distributions need 2 types: one integral for the discrete sample space, used in the generation of random values, and one fractional for the calculations of pdf, cdf etc. Changes also propagate to Distribution definition.
Open
|
Grouped in its own issue. This should probably go in a separate module, maybe with the spire.random subpackage. |
|
I will try to. Maybe I will need some help. |
|
Are you at nescala btw? If you do anything, please base your PR on the modularization PR, and put stuff in a separate module such as spire-stats or spire-random. We hope to get this big changes done as soon as possible, but we may hit roadblocks. |
JoeWrightss
pushed a commit
to JoeWrightss/spire
that referenced
this pull request
Apr 13, 2019
Merge AWS node resolver
|
Closing this because it hasn't seen activity in over a year. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hello again,
This is a second attempt to an alternate proposal for the random package as seen in #214. For more information about the first one (to avoid redundancy) refer to #460. The only difference is that this one complies with the restructuring that took place to add scala-js support.
I'm sorry that I cannot give continuous contributions to the project.
As always, feel free to suggest modifications and improvements.