[FIXED] Getting error while passing Class_weight parameter in Random Forest

Issue

I am doing binary classifier. Since my data is unbalanced i am using class weight. I am getting error while passing values how to fix this.

Error: ValueError: class_weight must be dict, 'balanced', or None, got: [{0: 0.4, 1: 0.6}]"

Code

rf=RandomForestClassifier(n_estimators=1000,oob_score=True,min_samples_leaf=500,class_weight=[{0:.4, 1:.6}])
    fit_rf=rf.fit(X_train_res,y_train_res)

Error

\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\utils\class_weight.py in compute_class_weight(class_weight, classes, y)
     60         if not isinstance(class_weight, dict):
     61             raise ValueError("class_weight must be dict, 'balanced', or None,"
---> 62                              " got: %r" % class_weight)
     63         for c in class_weight:
     64             i = np.searchsorted(classes, c)

ValueError: class_weight must be dict, 'balanced', or None, got: [{0: 0.4, 1: 0.6}]

How to fix this.

Solution

Per the documentation

class_weight : dict, list of dicts, “balanced”,

Therefore, the class_weight paramter accepts a dictionary, a list of dictionary, or the string "balanced". The error message you are given states that it wants a dictionary, and since you have only one dictionary a list is not needed.

So, let's try:

rf=RandomForestClassifier(n_estimators=1000,
                          oob_score=True,
                          min_samples_leaf=500,
                          class_weight={0:.4, 1:.6})

fit_rf=rf.fit(X_train_res,y_train_res)

Answered By - Scott Boston