combust/mleap

Reenable LDA parity spec

Open

#169 opened on Apr 6, 2017

View on GitHub
 (0 comments) (0 reactions) (0 assignees)Scala (1,461 stars) (312 forks)batch import
bughelp wanted

Description

LDA uses a random sampling to score data. This causes slightly different values to to be generated between each run, which makes testing outright equality of tensors a no-go for the parity spec.

We are temporarily disabling the LDA parity spec in favor of fixing tensor equality and will reenable the parity spec using lighter equality restrictions at a later time before the 0.7.0 release.

See the tensor equality issue here: https://github.com/combust/mleap/pull/156

Contributor guide

Tech stack
scala
Domain
machine learningtesting
Issue type
test
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
2
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
ScalaMLeap basicstesting
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
65
Research direction
Review PR #156 for the tensor equality fix. Locate the LDA parity spec file (likely LdaParitySpec.scala) and modify the test assertions to use approximate equality instead of exact tensor equality. Consider using a tolerance based comparison to account for random sampling variation. Re enable the spec before the 0.7.0 release.