I encountered an issue while using mljar-supervised with a boolean series as target. The produced predictions are not in [False, True] but instead [0, 1].
import numpy as np
import pandas as pd
x, y = pd.DataFrame(np.random.random(size=(150, 4))), pd.Series([True] * 75 + [False] * 75)
import supervised
from supervised.automl import AutoML
automl = AutoML(ml_taks="binary_classification", total_time_limit=60)
automl.fit(x, y)
automl.predict(x)
As you can see, the predictions produced by mljar-supervised are not in the original domain.
This also happens with predict_all's label column, the probability columns are called prediction_0 and prediction_1 accordingly.
I encountered an issue while using
mljar-supervised
with aboolean series
as target. The produced predictions are not in[False, True]
but instead[0, 1]
.As you can see, the predictions produced by
mljar-supervised
are not in the original domain. This also happens withpredict_all
's label column, the probability columns are calledprediction_0
andprediction_1
accordingly.edit: tested with
0.10.3
and0.10.6