how to get confusion matrix in form of matrix and roc curve as graph

User 3238 | 2/22/2016, 8:23:44 AM

I am getting confusion matrix in the tabular form instead of matrix is there is a way to show this in form of matrix & if there is not a way do this then how can i print all the rows. print_rows() function is not working as the type of object is 'dict'. i also tried to convert this object to sframe and print it but again not getting expected result.

Also how to get graph of roc curve.

Comments

User 19 | 2/22/2016, 6:01:43 PM

Hi Bhanu,

The evaluate method returned a dictionary (i.e. 'dict') where the values are SFrames. To get the confusion matrix and print all rows, you can do

result = kdd_model.evaluate(test_data, metric='confusion_matrix') conf_mat = result['confusion_matrix'] conf_mat.print_rows(200)

Similarly, your second command gave you all the data needed to plot an ROC curve.

result = kdd_model.evaluate(test_data, metric='roc_curve') roc_curve = result['roc_curve']

Now you just need to use your plotting tool of choice to plot the ROC curve, or if you've run the above code you may then call

kdd_model.show()

Let me know if that helps! Chris


User 3238 | 2/24/2016, 1:48:13 PM

Thanks ChrisDuBois,

I get the result i was expecting