RESULT ANALYSIS

PRESENTED BY
SafeNet AI

SAFENET AI
 CONTENTS
• Confusion Metrics
• Visualization Libraries
• Basic Visualization
• Advance & Interactive Visualization
• Tools
• Best Practices
• resources

SAFENET AI
 CONFUSION METRICS
Confusion metrics are a way to evaluate the performance of a
classification model by comparing its predictions with the
actual class labels.

SAFENET AI
 CONFUSION METRICS
 Components of Confusion Metrix:

 True Positives (TP): Instances where the model correctly predicts the
positive class.
 False Positives (FP): Instances where the model incorrectly predicts the
positive class.
 True Negatives (TN): Instances where the model correctly predicts the
negative class. SAFENET AI
 CONFUSION METRICS
A binary classification problem predicting whether an email is spam or not
using a simple dataset.

Actual / Predicted Spam (Positive) Not Spam

(Negative)
Spam (Positive) 20 (TP) 5 (FP)
Not Spam (Negative) 3 (FN) 72 (TN)
Example Dataset

 Interpretation:
 True Positives (TP): The model correctly identified 20 spam emails.
 True Negatives (TN): The model correctly identified 72 non-spam emails.
 False Positives (FP): The model incorrectly classified 5 non-spam emails
as spam.
 False Negatives (FN): The model missed 3 actual spam emails.

SAFENET AI
 ACCURACY
Accuracy is a fundamental metric in machine learning that
measures the overall correctness of a model's predictions.

Formula:

Accuracy:

SAFENET AI
 ACCURACY
A binary classification problem of predicting whether an image
contains a cat or not. The confusion matrix for our model is as
follows
Actual / Predicted Cat(Positive) Not Cat(Negative)
Cat (Positive) 85 (TP) 10 (FP)
Not Cat (Negative) 5 (FN) 900 (TN)

 Interpretation:
 True Positives (TP): The model correctly identified 85 images containing cats.
 True Negatives (TN): The model correctly identified 900 images without cats.
 False Positives (FP): The model incorrectly classified 10 images without cats as
containing cats.
 False Negatives (FN): The model missed 5 images containing cats.
 Calculation:
 Accuracy: (85 + 900) / (85 + 10 + 5 + 900) = 0.976

SAFENET AI
 PRECISION
Precision is a metric in machine learning that measures the
accuracy of positive predictions made by a model.

Formula:

Precision:

SAFENET AI
 PRECISION
A binary classification problem of predicting whether an email
is spam or not.
Actual / Predicted Spam (Positive) Not Spam
(Negative)
Spam (Positive) 120 (TP) 15 (FP)
Not Spam (Negative) 10 (FN) 850 (TN)
 Interpretation:
 True Positives (TP): The model correctly identified 120 spam emails.
 False Positives (FP): The model incorrectly classified 15 non-spam emails as
spam.
 True Negatives (TN): The model correctly identified 850 non-spam emails.
 False Negatives (FN): The model missed 10 actual spam emails.
 Calculation:
Precision: 120 / (120 + 15) = 0.889

SAFENET AI
 RECALL
Recall, also known as sensitivity or true positive rate, is a
metric in machine learning that measures the ability of a model
to capture all the relevant instances of the positive class.

Formula:

Recall (Sensitivity):

SAFENET AI
 RECALL
A binary classification problem of predicting whether a medical
test correctly identifies a disease.
Actual / Predicted Disease (Positive) No
Disease(Negative)
Disease (Positive) 90 (TP) 15 (FN)
No Disease 5 (FP) 890 (TN)
 (Negative)
Interpretation:
 True Positives (TP): The model correctly identified 90 instances of the disease.
 False Negatives (FN): The model missed 15 instances of the disease.
 True Negatives (TN): The model correctly identified 890 instances without the
disease.
 False Positives (FP): The model incorrectly classified 5 instances without the
 disease
Calculation:
as positive.
Recall (Sensitivity): 90 / (90 + 15) = 0.857

SAFENET AI
 F1-SCORE
The F1 Score is a metric in machine learning that combines
precision and recall into a single value, providing a balance
between the two.

Formula:

F1 Score:

SAFENET AI
 F1-SCORE
A binary classification problem of identifying whether a
transaction is fraudulent
Actual / Predicted Fraud(Positive) Not Fraud
(Negative)
Disease (Positive) 70 (TP) 15 (FN)

 Not Fraud (Negative)

Interpretation:
5 (FP) 900 (TN)

 True Positives (TP): The model correctly identified 70 instances of fraudulent

transactions.
 False Positives (FP): The model incorrectly classified 15 non-fraudulent
transactions as fraudulent.
 True Negatives (TN): The model correctly identified 900 non-fraudulent
transactions.
 Calculation:
 False Negatives (FN): The model missed 5 instances of fraudulent transactions.
Precision: 70 / (70 + 15) = 0.824
Recall (Sensitivity): 70 / (70 + 5) = 0.933

 F1-Score2 * (0.824 * 0.933) / (0.824 + 0.933) = 0.857

SAFENET AI
 RESOURCES

• https://www.analyticsvidhya.com/blog/2020/04/confusion-matrix-machine-lea

rning/

• https://towardsdatascience.com/understanding-confusion-matrix-a9ad42dcfd

62

SAFENET AI