Update final_report.md

tabs/final_report.md

@@ -202,7 +202,7 @@ We fit our decision tree with the cost complexity hyperparameter described [prev
 
 <img src="../assets/dt_confusion_matrix.png" alt="drawing" width="300"/>
 
-**A note on F1-Score and AUC:** For this section, we use a weighted average F1-score and weighted average Area Under the receiver operating Curve (AUC). The reason we weight these scores is due to the imbalance in the classes for this dataset. The F1-score metric is the harmonic mean of precision and reall. Thus, it acts as an aggregated metric for both precision and recall. Because it's defined on a binary case of true/false postive/negatives, each class has its corresponding F1-score. These values are then aggregated by a weighted average into one value, as reported below. The AUC metric is an aggregate measurement of true and false positive rates derived from the ROC plot, which plots the true positive rate (TPR) against the false positive rate (FPR) at each threshold setting. Similarly to the F1-score, this is a binary classification statistics. Therefore, each class has their own AUC score which is aggregated into a single reported AUC. We use both the 1 vs Rest and 1 vs 1 methods. 1 vs Rest divides the data into two classes as the 1 class we are measuring (positive), and the rest (negatives). The 1 vs 1 method only looks at pairwise comparisons between each class as the positives and negatives. Both of the metrics for measuring classification performance are highly regarded and tend to perform better than accuracy alone, especially in imbalanced datasets such as this one [[5.]](#references), [[6.]].
+**A note on F1-Score and AUC:** For this section, we use a weighted average F1-score and weighted average Area Under the receiver operating Curve (AUC). The reason we weight these scores is due to the imbalance in the classes for this dataset. The F1-score metric is the harmonic mean of precision and reall. Thus, it acts as an aggregated metric for both precision and recall. Because it's defined on a binary case of true/false postive/negatives, each class has its corresponding F1-score. These values are then aggregated by a weighted average into one value, as reported below. The AUC metric is an aggregate measurement of true and false positive rates derived from the ROC plot, which plots the true positive rate (TPR) against the false positive rate (FPR) at each threshold setting. Similarly to the F1-score, this is a binary classification statistics. Therefore, each class has their own AUC score which is aggregated into a single reported AUC. We use both the 1 vs Rest and 1 vs 1 methods. 1 vs Rest divides the data into two classes as the 1 class we are measuring (positive), and the rest (negatives). The 1 vs 1 method only looks at pairwise comparisons between each class as the positives and negatives. Both of the metrics for measuring classification performance are highly regarded and tend to perform better than accuracy alone, especially in imbalanced datasets such as this one [[5.]](#references), [[6.]](#references).
 
 Decision Tree Classifier Results:
 - Training Accuracy: 1.0