Exam AI-900 topic 1 question 3 discussion

Answer 2 is also correct please refer the https://docs.microsoft.com/en-us/azure/machine-learning/classic/evaluate-model-performance

Answer 2 is correct. The grid used in the question is reversed to the MS documentation. A false negative would imply that a 0 was predicted but 1 was the actual outcome, of which there were 1033 occurrences according to the grid used in the question.

In the confusion matrix, the cell where the prediction is 1 (positive) and the actual value is also 1 represents the true positives. In this exhibit, that value is 11, meaning the model correctly predicted 11 positive cases. Conversely, the cell where the model predicted 0 (negative) but the actual value was 1 represents the false negatives.

To evaluate how well the model is performing, we use a confusion matrix. This matrix summarizes how many times the model made each type of prediction. It categorizes predictions into four groups: True Positives (TP): These are cases where the model correctly predicts a positive outcome. For example, if an email is spam, and the model correctly identifies it as spam. True Negatives (TN): These are cases where the model correctly predicts a negative outcome. For example, if an email is not spam, and the model correctly identifies it as not spam. False Positives (FP): These are cases where the model incorrectly predicts a positive outcome. For example, if an email is not spam, but the model mistakenly labels it as spam. This is also called a "Type I error." False Negatives (FN): These are cases where the model incorrectly predicts a negative outcome. For example, if an email is spam, but the model mistakenly labels it as not spam. This is also called a "Type II error."

Actual: 1 0 Predicted: 1 11 5 0 1033 13951 X-axis (top): Represents the Actual values. 1: Represents cases where the actual label is 1 (positive). 0: Represents cases where the actual label is 0 (negative). Y-axis (left): Represents the Predicted values. 1: Represents cases where the model predicted 1 (positive). 0: Represents cases where the model predicted 0 (negative). This matrix shows: True Positives (TP): 11 (Predicted 1, Actual 1) – Model correctly predicted "1". False Positives (FP): 5 (Predicted 1, Actual 0) – Model predicted "1" when it should have been "0". False Negatives (FN): 1033 (Predicted 0, Actual 1) – Model predicted "0" when it should have been "1". True Negatives (TN): 13951 (Predicted 0, Actual 0) – Model correctly predicted "0".

The confusion matrix for the model scored on test data reveals the following: - **True Positives (Correctly Predicted Positives)**: 11 - **False Negatives**: 5 This information is crucial for evaluating the model's performance, especially in a classification task. The true positives represent instances where the model correctly predicted the positive class, while false negatives indicate instances where the model failed to predict the positive class when it should have. Understanding these metrics helps assess the model's accuracy and effectiveness in identifying positive cases. Source: Conversation with Copilot, 27/6/2024

No matter the axis labels (X and Y), the positions of TP, FP, FN, and TN remain consistent within the matrix. Actual Predicted 1 0 1 11 5 0 1033 13951 Interpretation: True Positives (TP): Predicted 1, Actual 1 — Value: 11 False Positives (FP): Predicted 1, Actual 0 — Value: 5 False Negatives (FN): Predicted 0, Actual 1 — Value: 1033 True Negatives (TN): Predicted 0, Actual 0 — Value: 13951

In Question, Actual is on X-axis and Predicted is on 'Y' axis So answer will be 11 and 1033 TP = 11 TN = 13951 FP = 5 FN = 1033 In Solution, Actual Values are on 'Y-axis' and Predicted Values on 'X-axis' - Thus table get changed.

Hello team, I do not understand this can someone explain why the answer is 11:1033? I do not understand the logic. Thank you

I had a lot of difficulties with the answer for this question. However, now that I have been able to step back and look at the chart objectively, you realize that it is not a chart for determining, it a representational chart, where you are being presented the results. That is why the response is actually correct.

Explanation chart shown is wrong Should be TP FP FN TN

11 corresponds to 0-0 1033 corresponds to 0-1

Do not agree with the response for answer 2. If we have a binary classification 'let me say person is 'tall' or 'not tall' (two classes). The confusion matrix shows the top row as row 1 / column 1 which are the true positives (correctly identified as tall). The next value in top row: row 1 and column 0 that is defined as tall (row) and 0 (contains incorrectly classified as 'not tall') and is called false negative. I read the questions that 'false negative' is what they are asking for. table: tall: correctly predicted - not tall predicted not_tall: incorrectly classified as not tall - correctly identified as not tall.

I have found some answer are wrong and explanation is not proper. You can use this course for proper explanation and correct answers https://www.udemy.com/course/microsoft-azure-ai-900-latest-practice-test-2023/?referralCode=25EA856C8C63BC4DE950

1. 11 2. 1033 https://learn.microsoft.com/en-us/azure/machine-learning/how-to-understand-automated-ml?view=azureml-api-2#confusion-matrix

answers and explanations are correct

PP = 11 and FN = 1033