Page 143 - Ai_C10_Flipbook
P. 143

UNIT-3





                                                                EVALUATING MODELS













                 Evaluation is the key step in the AI project cycle where the effectiveness, accuracy, and reliability of the model are
                 assessed to ensure it meets the project objectives and performs well in real-world scenarios.
                 After the designing of an AI model, it is important to evaluate the model to see that the model is designed as per
                 the need and is giving the desired process.

                         Understanding Evaluation


                 It is the evaluation phase of the AI project cycle, where we check whether the model can achieve required goals or
                 not. If the model fails to meet the requirements, we can modify either the model or the data. Once the developer
                 feels it is deployed in a real-world environment and handed over to the user. If the deployment stage is not
                 reached, the project is of no use.

                 Importance of Model Evaluation

                 Model evaluation is the process of applying various metrics to assess a machine learning model’s performance.
                 AI model improves overtime with constructive feedback. This is an iterative process where you build the model,
                 evaluate its performance using appropriate metrics, refine it based on the feedback, and repeat until the desired
                 accuracy is achieved. It’s similar to tuning a musical instrument—regularly checking its sound quality, making
                 adjustments, and fine-tuning until the melody is harmonious and meets the desired standard.
                 Some of the advantages of evaluating a model are as follows:
                    • Evaluation ensures that the model is operating correctly and optimally.

                    • Evaluation is an initiative to understand how well it achieves its goals.
                    • Evaluation helps to determine what works well and what could be improved in a program.

                      Human
                                    You learn a            You take a          You assess the         You thrive for
                     Learning
                                      subject                 test                 result             better results
                      Process

                     Machine        Training the          Testing the                                Fine tuning the
                                                                               Evaluating the
                     Learning       model with            model with                                 model for better
                                                                                   model
                      Process      training data.         testing data                                performance
                 Need for Model Evaluation

                 Model  evaluation  is  like  giving  the  AI  model  a  performance  review.  It  helps  identify  its  strengths,  pinpoint
                 weaknesses, and determine how well it fits the task at hand. This feedback acts as a guide to refine and improve
                 the model, making it more reliable and trustworthy. The process is continuous, just like fine-tuning a skill to
                 achieve the best results. Depending upon the type and the purpose of the evaluation model, there are different
                 types of evaluation techniques, like Train-Test split, Confusion Matrix, etc.

                                                                                           Evaluating Models    141
   138   139   140   141   142   143   144   145   146   147   148