How To Plot A Roc Curve Using Dataframe Converted From Csv File

I was trying to plot a ROC curve by using the documentation provided by sklearn. My data is in a CSV file, and it looks like this.It has two classes 'Good'and 'Bad' screenshot of m

Solution 1:

Looks like you simply don't understand how your data is structured and how your code should work.

LabelBinarizer will return a one-v-all encoding, meaning that for two classes you will get the following mapping: ['good', 'bad', 'good'] -> [[1], [0], [1]], s.t. n_classes = 1.

Why would you expect it to be 3 if you have 2 classes? Simply change plt.plot(fpr[2], tpr[2], color='darkorange', lw=lw, label='ROC curve (area = %0.2f)' % roc_auc[2]) to plt.plot(fpr[0], tpr[0], color='darkorange', lw=lw, label='ROC curve (area = %0.2f)' % roc_auc[0]) and you should be good.

Solution 2:

Just look in tpr and fpr dictionaries and you will see that you don't have a tpr[2] or fpr[2]. n_classes = y.shape[1] shows how many classes you have (= 2), which means that you have keys of 0 and 1 in your tpr and fpr dictionaries.

You are overcomplicating things by using a multi-class approach when you only have 2 classes (binary classification). I think you are using this tutorial.

I would advise replacing the following:

# Compute ROC curve and ROC area for each class
fpr = dict()
tpr = dict()
roc_auc = dict()
for i inrange(n_classes):
    fpr[i], tpr[i], _ = roc_curve(y_test[:, i], y_score[:, i])
    roc_auc[i] = auc(fpr[i], tpr[i])

# Compute micro-average ROC curve and ROC area
fpr["micro"], tpr["micro"], _ = roc_curve(y_test.ravel(), y_score.ravel())
roc_auc["micro"] = auc(fpr["micro"], tpr["micro"])
plt.figure()
lw = 2
plt.plot(fpr[2], tpr[2], color='darkorange',
         lw=lw, label='ROC curve (area = %0.2f)' % roc_auc[2])

With something like:

fpr, tpr = roc_curve(y_test.values, y_score[:,1])
roc_auc = auc(fpr,tpr)
plt.figure()
lw = 2
plt.plot(fpr, tpr, color='darkorange', lw=lw, label='ROC curve (area = %0.2f)' % roc_auc)