Page 246 - AI Ver 3.0 class 10_Flipbook
P. 246

The term Error means the action that is inaccurate or wrong. It refers to the difference between a model’s prediction
                    and the actual outcome. It quantifies how often the model makes mistakes. Based on the error, we choose the machine
                    learning model that has the best performance for a specific dataset. Low Error in a model performance signifies precise
                    and reliable predictions.
                  5.  Discuss the ethical concerns around model evaluation.
                Ans.  The following are the ethical concerns in the model evaluation:
                     •   Bias: Evaluation metrics may fail to detect biases in a model, leading to unfair outcomes. For example, a model might
                       favor one gender, race, or socio-economic group over another. To prevent this, metrics should be carefully designed
                       to avoid introducing or perpetuating bias.
                     •   Fairness: Fairness ensures that models treat all groups equally. The evaluation processes must account for fairness
                       to prevent models from producing discriminatory results.
                     •   Transparency: Sometimes, evaluation models lack clarity, making it difficult to understand how predictions are made.
                       A transparent approach clearly explains how metrics are chosen and how results are derived.
                     •   Accountability: It is crucial to take responsibility for the choice of evaluation metrics and their outcomes. Providing
                       clear reasoning behind metric selection helps ensure ethical decision-making and builds trust.
                     •   Privacy: Using real-world data for evaluation often involves sensitive personal information, raising concerns about
                       privacy. Measures should be in place to ensure that individual data is protected during the evaluation process.
                     •  Data Protection: Protecting the data used in model evaluation is essential to prevent misuse or unauthorised access.

              C.  Competency-based/Application-based questions.                               21 st  Century   #Critical Thinking
                                                                                                 Skills
                  1.  An AI model made the following predictions for Book Sales forecast. Calculate Accuracy, precision and recall for the
                    following confusion matrix:

                          Confusion Matrix            True Positives            True Negatives

                          Predicted Positive               50                        40


                         Predicted Negative                12                        10

                                     Correction prediction
                Ans.  Accuracy =                         × 100%
                                         Total Cases
                                          (TP + TN)
                             =                           × 100%
                                      (TP + TN + FP + FN)
                                           50 + 10
                             =                           × 100%
                                       50 + 10 + 40 + 12
                                      60
                             =            × 100%
                                     112
                             =      53.5%
                                         True Positive
                    Precision =
                                     All Predicted Positive
                                       TP
                             =
                                     TP + FP
                                       50
                             =
                                     50 + 40

                             =      0.555 or 55.5%



                    244     Touchpad Artificial Intelligence (Ver. 3.0)-X
   241   242   243   244   245   246   247   248   249   250   251