predict not implemented

[jrgasper]

Predict is not implemented in class treefarms yet. Is there a suggested workaround?

TODO: implement this
def predict(self, X):
"""
Parameters
---
X : matrix-like, shape = [n_samples by m_features]
a matrix where each row is a sample to be predicted and each column is a feature to be used for prediction
Returns
---
array-like, shape = [n_sampels by 1] : a column where each element is the prediction associated with each row
"""
raise NotImplementedError

[thatswhatsimonsaid]

Hi there! I'm also interested in the predict. Did you find a workaround?

[thatswhatsimonsaid]

Hi there! I took a dive and wanted to offer some insight.

There is a predict and classify function in TreeClassifier. So for each model in the Rashomon Set, you can get easily get the prediction. In the tutorial, for instance, you can write first_tree.predict(df) to get the prediction labels from the first tree.

To get it for all models in the Rashomon Set, you can easily loop it:

PredictedValues = [model[i].predict(df) for i in range(model.get_tree_count())]

Note that the predict function in TreeClassifier uses a for loop, but it might be more efficient in large datasets to replace the loop by vectorizing as such:

predictions = X.apply(lambda row: self.classify(row.values)[0], axis=1)

I plan on submitting a pull request on this soon (it would be my first ever pull request :) ).