Getting inf% accuracy for Random Forest model with scikit learn

Question

I have pandas dataframe with 8000 observations and with many different columns -some of them are dates and hours that were converted into dummies (get_dummies) and some are numerical,and y label which is what I want to predict that has two classes: 0 or 1:

>>>y   x1     x2     x3     x4   10   11   12   13
0  1  0.532  0.431  0.214  0.11  0    0    1    0
1  0  0.512  0.410  0.340  0.09  0    1    0    0
...

I have used random forest model in scikit learn in order to predict the y column (treatment) either is 1 or 0.

the results suprised me because when I print the accuracy I got -inf and I did not excpect that.

this is how I calcualte and run the rf:

X = df.drop(['y',],axis=1)
y = df['y'].astype(int)


#split the data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=123)

###Random forest

#function for evaluate:
def evaluate(model, test_features, test_labels):
    predictions = model.predict(test_features)
    errors = abs(predictions - test_labels)
    mape = 100 * np.mean(errors / test_labels)
    accuracy = 100 - mape
    print('Model Performance')
    print('Average Error: {:0.4f} degrees.'.format(np.mean(errors)))
    print('Accuracy = {:0.2f}%.'.format(accuracy))
    
    return accuracy

#random forest
from sklearn.ensemble import RandomForestClassifier
rfc = RandomForestClassifier()
rfc.fit(X_train,y_train)
predictions = rfc.predict(X_test)

base_model = RandomForestClassifier()
base_model.fit(X_train, y_train)
base_accuracy = evaluate(base_model, X_test, y_test)

>>>Model Performance
Average Error: 0.3290 degrees.
Accuracy = -inf%.

the confusion matrix looks like this:

so my question is what that -inf% accuracy mean? where is the problem in the model?

Answer 1

Your implementation of accuracy is incorrect:

predictions = model.predict(test_features)
errors = abs(predictions - test_labels)
mape = 100 * np.mean(errors / test_labels)
accuracy = 100 - mape

You should use scikit-learn's accuracy function:

Something like:

from sklearn.metrics import accuracy_score

y_pred = model.predict(test_features)
accuracy = accuracy_score(test_labels, y_pred)