How to persist DTree? about ml_algo

[gyrdym]

@mohas Hi, thank you for opening the issue, I will add the desired functionality to the classifier as soon as possible - it's one of the things that I planned to do first. Unfortunately, right now you can't save the trained model. Just keep tracking this issue, I'll write when the functionality is ready.

[mohas]

Coming from other programming languages I thought I could just binary serialize the dtree class itself, but apparently this is not easily doable in dart

[gyrdym]

@mohas sorry for such a big delay, finally I've added the desired functionality to DecisionTreeClassifier. You can use the following example as a guide:

final fakeDataSet = DataFrame.fromSeries([
  Series('col_1', <int>[10, 90, 23, 55]),
  Series('col_2', <int>[20, 51, 40, 10]),
  Series('col_3', <int>[1, 0, 0, 1], isDiscrete: true),
  Series('col_4', <int>[0, 0, 1, 0], isDiscrete: true),
  Series('col_5', <int>[0, 1, 0, 0], isDiscrete: true),
  Series('col_6', <int>[30, 34, 90, 22]),
  Series('col_7', <int>[40, 31, 50, 80]),
  Series('col_8', <int>[0, 0, 1, 2], isDiscrete: true),
]);

final targetName = 'col_8';

final classifier = DecisionTreeClassifier(fakeDataSet, targetName, minError: 0.3, minSamplesCount: 1, maxDepth: 3);
final testFileName = 'test/classifier/decision_tree_classifier/serialized_classifier.json';
await classifier.saveAsJson(testFileName);

And to restore the classifier:

final file = await File(testFileName);
final json = await file.readAsString();
final restoredClassifier = DecisionTreeClassifier.fromJson(json);

You may visit the appropriate unit test to see it in action.