Conversion of Multi Class Naive Bayes classifier to estimator #1111
Conversation
src/Microsoft.ML.StandardLearners/Standard/MultiClass/MultiClassNaiveBayesTrainer.cs
Show resolved
Hide resolved
| /// <summary> | ||
| /// Initializes a new instance of <see cref="MultiClassNaiveBayesTrainer"/> | ||
| /// </summary> | ||
| /// <param name="env">The private instance of <see cref="IHostEnvironment"/>.</param> |
There was a problem hiding this comment.
The private instance of . [](start = 30, length = 55)
this is not helping.
The environment to use. #Resolved
| .Concat(MetadataUtils.GetTrainerOutputMetadata())); | ||
| return new[] | ||
| { | ||
| new SchemaShape.Column(DefaultColumnNames.Score, SchemaShape.Column.VectorKind.Vector, NumberType.R4, false, new SchemaShape(MetadataForScoreColumn())), |
There was a problem hiding this comment.
MetadataForScoreColumn [](start = 141, length = 22)
- why the disparity between this and the line below, where you initialize
metadataright in the method? - method names should be verbs. So
GetScoreColumnMetadataor something #Resolved
| var reader = TextLoader.CreateReader(env, | ||
| c => (label: c.LoadText(0), features: c.LoadFloat(1, 4))); | ||
|
|
||
| MultiClassNaiveBayesPredictor pred = null; |
There was a problem hiding this comment.
MultiClassNaiveBayesPredictor [](start = 12, length = 29)
Should we expose Label and Feature histograms? Otherwise, I don't see any reason why we actually expose it. #Closed
There was a problem hiding this comment.
Created two public methods that will output a copy of the featurehistogram and labelhistogram
In reply to: 222122921 [](ancestors = 222122921)
| using Microsoft.ML.Runtime.Internal.Calibration; | ||
| using Microsoft.ML.Runtime.Learners; | ||
| using Microsoft.ML.StaticPipe.Runtime; | ||
|
|
There was a problem hiding this comment.
change file name to static #Resolved
| public void GetLabelHistogram(ref int[] labelHistogram, out int labelCount) | ||
| { | ||
| labelCount = _labelCount; | ||
| Utils.EnsureSize(ref labelHistogram, _labelCount, _labelCount); |
There was a problem hiding this comment.
_labelCount, _labelCount [](start = 49, length = 24)
Why repeat twice?
Why repeat twice?
We have method which accept only one value, or you deliberately want to shrink array even if it slighlty bigger? #Closed
There was a problem hiding this comment.
I was thinking of returning exactly the right size for the array, but I see why if we pass a bigger array it might be better not to shrink it.
In reply to: 222538588 [](ancestors = 222538588)
ghost
locked as resolved and limited conversation to collaborators
Ongoing work on converting the trainers to estimators (#754). This PR converts the Multi Class Naive Bayes classification trainer.