# Model Evaluation Metrics

Before going into the details, you must be known for what we are going to discuss. So, this series will see some model evaluation techniques. Let’s get started. Instead of directly jumping to the methods of model evaluation, let’s recap the basic process involved in machine learning statistical models.

The basic machine learning approach is like data pre-processing, building statistical model, performance or error measures and labeling the unseen data. So, many newbies have done a mistake of directly apply the build model on unseen data which is not a good at all. As the first one need to evaluate its model. Our statistical model can be Classification model or Regression model, both regression model and classification model has different evaluation metrics.

Hence, in upcoming blogs will cover these evaluation metrics:

- Confusion Matrix
- Accuracy
- Precision and Recall
- F1 Score
- ROC – Area Under Curve
- Gini Coefficient
- Least Square Error
- Mean Absolute Error
- Root Mean Square Error

## 1. Confusion Matrix

Isn’t name interesting? Let’s dive into the details of it. Confusion Matrix is generally used for classification problems. So classification problem can give binary output or multi-class output as well then Confusion Matrix is a kind of square matrix. If problems consist N classes then Confusion Matrix will have dimensions N*N. Here for simplicity, considering 2 classes(True and False).

Below is the Confusion Matrix for True and False.

There are four basic terms involved to be discussed-:

**a.) True Positive – **This term states that Actual Label is Positive and Predicted Label is also Positive, hence stated as True Positive.

**a.) False Positive – **This term states that Actual Label is Positive and Predicted Label is Negative, hence our model is predicting true for false values too. Therefore, stated as False Positive

**a.) True Negative – **This term states that Actual Label is Negative and Predicted Label is also Negative, hence stated as True Negative.

**a.) False Negative – **This term states that Actual Label is Negative and Predicted Label is Positive, hence stated as True Positive.

By using these basic terms will derive precision, recall, and accuracy for any model. This confusion matrix is very helpful as everything gets into the mind so easily by this diagram. So, this is all for this tutorial.

Hope you understand the Confusion Matrix. In the next blog, will going to discuss Accuracy i.e, how to evaluate the accuracy of the model.