Page 151 - Ai_C10_Flipbook
P. 151
Steps to Fill in the Confusion Matrix
• Count the number of rows having Yes in both the columns of the Predicted
Confusion Matrix
table and put the count of it in the top left cell. Yes No
• Similarly, the number of rows having Yes in the Actual Value Yes 3 2
column and No in the Predicted Value column will be shown in Actual
No 2 3
the top right cell of the confusion matrix.
• Number of rows having No in the Actual Value and Yes in the Predicted Value column will be shown in the
down left cell of the confusion matrix.
• Lastly, the number of rows having No in the both columns will be shown in the down right cell of confusion
matrix.
So, the final Confusion matrix will be as follows:
Here, the total number of correct predictions are 6 out of 10
The classification outcomes based on the different values of actual and predicted labels are as follows:
• True Positive
• True Negative
• False Positive
• False Negative
True Positive
A True Positive occurs when a model correctly predicts a positive
Predicted
outcome. In the above example, the value of the True Positive is Confusion Matrix
depicted in the shaded region. Yes No
Some more examples of True Positive are as follows: Yes 3 2
Actual
• Medical Diagnosis - A machine learning model predicts
No 2 3
whether the patient has asthma.
True Positive: The model predicts that the patient actually has asthma.
• Face Recognition Security System - A security system identifies individuals who are authorised to enter a
restricted area.
True Positive: The system recognises an authorized employee correctly.
True Negative
A True Negative occurs when a model correctly predicts a
Predicted
negative outcome. When the model’s negative prediction is same Confusion Matrix
as the actual outcome, it’s the case of True Negative. In the given Yes No
example the shaded region depicts the True Negative scenario. Yes 3 2
Actual
Some more examples of True Negative are as follows:
No 2 3
• Spam Detection - The model predicts an email is “Not Spam”.
True Negative is when an email is indeed not spam.
• Loan Default Prediction - The model predicts a customer will not default, in payment of loan instalment on time.
True Negative is the customer paid and did not default.
Evaluating Models 149

