Model Evaluation

Answer

Model evaluation is the process of understanding the reliability of any AI model by feeding the test dataset into the model and comparing its outputs with the actual answers. It shows how well the mod…

Answer

Testing data should be used because using training data can make the model seem better than it really is. The model may simply remember the training set, which leads to overfitting. Testing data gives…

Answer

Accuracy is the percentage of correct predictions out of the total number of predictions. Formula: Accuracy = (Number of Correct Predictions) / (Total Number of Predictions). It is useful, but it can …

Answer

Error shows the proportion of wrong predictions. Formula: Error = 1 - Accuracy. If accuracy increases, error decreases.

Answer

High accuracy can be misleading in class-imbalanced datasets. For example, if a model always predicts no fire in a forest fire scenario, it can still get 98% accuracy when fires are rare. Even then, t…

Answer

A confusion matrix is a table with 4 combinations of predicted and actual values: TP, TN, FP, and FN. It is not an evaluation metric itself, but a record that helps in evaluation.

Answer

True Positive means Prediction = Yes and Reality = Yes. It means the prediction matches the real situation. Example: The model predicts fire, and fire is actually present.

Answer

True Negative means Prediction = No and Reality = No. It means the prediction matches the real situation. Example: The model predicts no fire, and there is actually no fire.