通过 cross_val 保存 sklearn 分类器拟合

Question

I've got a classifier I'm fitting using a cross_val and getting good results. Essentially all I'm doing is:

clf = RandomForestClassifier(class_weight="balanced")
scores = cross_val_score(clf, data, target, cv=8)
predict_RF = cross_val_predict(clf, data, target, cv=8)

from sklearn.externals import joblib
joblib.dump(clf, 'churnModel.pkl')

Essentially what I want to do is take the model that's getting fit by cross_val and export to joblib. However when I try to pull it in in a separate project I get:

sklearn.exceptions.NotFittedError: Estimator not fitted, call `fit` before exploiting the model.

So I'm guessing cross_val is not actually saving the fit to my clf? How do I persist the model fit that cross_val is generating?

Answer 1

juanpa.arrivillaga is right. I am afraid you would have to do it manually, but scikit-learn makes it quite easy. The cross_val_score create trained models that are not returned to you. Below you would have the trained models in a list (ie clf_models)

from sklearn.model_selection import StratifiedKFold
from sklearn.ensemble import RandomForestClassifier
from copy import deepcopy

kf = StratifiedKFold(n_splits=8)
clf = RandomForestClassifier(class_weight="balanced")
clf_models = []

# keep in mind your X and y should be indexed same here
kf.get_n_splits(X_data)
for train_index, test_index in kf.split(X_data, y_data):
    print("TRAIN:", train_index, "TEST:", test_index)
    X_train, X_test = X_data[train_index], X_data[test_index]
    y_train, y_test = y_data[train_index], y_data[test_index]
    tmp_clf = deepcopy(clf)
    tmp_clf.fit(X_train, y_train)

    print("Got a score of {}".format(tmp_clf.score(X_test, y_test)))
    clf_models.append(tmp_clf)

-edit via juanpa.arrivillaga's advice StratifiedKFold is a better choice. Here I selected just for demonstration purposes.