Need more details

[mairaw]

@abidur commented on Wed May 23 2018

It would be nice to have more details on this beyond, "Number of weak hypotheses in the ensemble"

---

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: 12f802b4-7e85-f280-316a-8c0f026f7276
• Version Independent ID: a6ae94b0-5d3e-0d89-1c2b-802b55f79a1b
• Content: FastTreeBinaryClassifier.NumTrees Property (Microsoft.ML.Trainers)
• Content Source: dotnet/xml/Microsoft.ML.Trainers/FastTreeBinaryClassifier.xml
• Service: unspecified
• Product: unspecified
• GitHub Login: @dotnet-bot

[shauheen]

Thanks for filling, I think this is somewhat related to #160. We will publish more documentation as we go.

[justinormont]

This string originates from:

machinelearning/src/Microsoft.ML.FastTree/FastTreeArguments.cs

Line 241 in c023727

[Argument(ArgumentType.LastOccurenceWins, HelpText = "Number of weak hypotheses in the ensemble", ShortName = "iter", SortOrder = 1)]

We should modify it to say something more user friendly. Perhaps, "Total number of boosted trees to create in the ensemble".

@codemzs: Is this parameter (and string value) inherited by other learners, like FastForest? The current wording seems oddly non-specific to boosted trees.

[mitevpi]

I couldn't find any references to the parameter in other learners. Also judging by the documentation of the arguments around it, it seems like this may have just missed a legibility update.

[shauheen]

Thanks @jwood803 for #372 . @mairaw do you think we can close this issue now?

[mairaw]

@abidur was the one who originally opened this issue in our repo.

[shauheen]

Closing the issue, please file a new issue if the enhancements are not enough.