CODEWITHSUNDEEP
pythonmachine-learningscikit-learnclassificationmultilabel-classification
yassine sfayhi
yassine sfayhi

Reputation: 43

sklearn Cross validation scoring , scores are all nan

I'm trying to make a multiclass classification here and the score from the cross validaiton are all nan

Below the code which works perfectly for binary classifcation

when i only keep accuracy and balanced_accuracy it shows the actual score when i add f1 or precison or recall all scores turns into nan

the problem that my code worked perfectly for binary classifcation and i'm using the same dataset just changed the target data

scoring = {'accuracy': 'accuracy',
          "balanced_accuracy": "balanced_accuracy",
         "precision": "precision",
          "recall": "recall",
          "f1" :"f1",
          "roc_auc":"roc_auc" }
# load the dataset
def load_dataset(df):
    # load the dataset as a numpy array
    data = df
    # retrieve numpy array
    data = data.values
    # split into input and output elements
    X, y = data[:, :-1], data[:, -1]
    y = LabelEncoder().fit_transform(y)
    return X, y
 
# evaluate a model
def evaluate_model(X, y, model):
    # define evaluation procedure
    cv = RepeatedStratifiedKFold(n_splits=10, n_repeats=3, random_state=1)
    # evaluate model
    scores = cross_validate(model, X, y, scoring=scoring, cv=cv, n_jobs=-1)
    return scores
 

model=DecisionTreeClassifier()
 
# define the location of the dataset

# load the dataset
X, y = load_dataset(df2)

# evaluate the model and store results
results_without_nlp = evaluate_model(X, y, model)

i have tried to use those

from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score but that does not seem to help

Upvotes: 0

Views: 456

Answers (1)

trandangtrungduc
trandangtrungduc

Reputation: 335

For precision_score, recall_score and f1_score, I think you can try using parameter average = micro (or macro and weighted) for multiple targets. Because its default value is binary.

Upvotes: 1

Related Questions