Hello, my code is nearly the same as the correction but i get this error
entry 8 is
from sklearn.model_selection import validation_curve
import numpy as np
range_gamma = np.logspace(-3, 2, num=30)
train_scores, test_scores = validation_curve(
clf, data, target, param_name="svc__gamma", param_range=range_gamma,
cv=ShuffleSplit(random_state=0), scoring="neg_mean_absolute_error", n_jobs=2)
clf is the pipeline
clf = make_pipeline(StandardScaler(), SVC())
I think you do not need to provide a scoring parameter in your “validation_curve” as the score is already calculated between the target_predicted and the target_test.
We used it in the lecture because our prediction was linear and not categorical hence we needed to compute our way out to evaluate the score.
The real issue is that you are using a classifier (SVC()) and a regression metric (neg_mean_absolute_error). The error given by scikit-learn is not super informative thought.
From the error, I assume that you try to solve a classification problem. The target contains the classes: donated and not donated. Thus you need to use a classification metric when running the validation curve: 3.3. Metrics and scoring: quantifying the quality of predictions — scikit-learn 0.24.2 documentation
Probably accuracy would be enough and as mentioned by @nineggs, it would be this metric used if you don’t specify scoring.
Thank you, that’s it
Thank you Guillaume, with this error i understand the differences between metric and categorical scoring errors, that i did not understood before…