Question

Evaluate the performance of the Machine Learning-Deep learning model and discuss how the choice of input features affects the model performance.

Answer 1

The performance of the Machine Learning-Deep learning model is evaluated using the following metrics

1. Confusion matrix
2. Recall
3. Precision
4. Accuracy
5. Specificity
6. Precision-Recall
7. F1 score
8. ROC (Receiver Operating Characteristics)

Feature selection is one of the most important concepts of machine learning. It has huge impact on model performance and accuracy.

Data cleaning and feature selection must be done in the very initial stages of developing a model. If you proceed to model without data cleaning and feature selection process the accuracy of the model will be drastically reduced.

Some of the advantages of performing the above operations

1. improves accuracy
2. reducing training time
3. reduces overfitting

for example,

you are developing a model for predicting the salary of an employee we may consider features like age, position, experience etc..

but salary has nothing to do with employee name or his family details so we need to remove (not to select) this type of features in feature selection