ConfusionMatrix - smart1004/doc GitHub Wiki

Convert scikit-learn confusion matrix to pandas DataFrame ***** https://gist.github.com/nickynicolson/202fe765c99af49acb20ea9f77b6255e

@ Pretty print for sklearn confusion matrix
https://gist.github.com/zachguo/10296432

=====

def confusion_matrix(y_true=None, y_pred=None, labels=None): ''' Dataframe of confusion matrix. Rows are actual, and columns are predicted.

Parameters
----------
y_true : array
y_pred : array
labels : list-like

Returns
-------
confusion_matrix : DataFrame
'''
df = (pd.DataFrame(metrics.confusion_matrix(y_true, y_pred),
                   index=labels, columns=labels)
        .rename_axis("actual")
        .rename_axis("predicted", axis=1))
return df 

=========== 멀티 클래스
https://docs.aws.amazon.com/ko_kr/machine-learning/latest/dg/multiclass-classification.html

https://docs.aws.amazon.com/ko_kr/machine-learning/latest/dg/machinelearning-dg.pdf#about-batch-predictions

3개의 클래스에 대해 설명을 하고 있다 http://nate9389.tistory.com/1138

2.3. 혼동 행렬 (confusion matrix) 2.3.1. 7은 C1으로 분류돼야 할 문제에서 C1으로 예측한 횟수를 나타냄 2.3.2. 정확도 (Accuracy) := (7+8+9) ÷ (1+2+3+4+5+6+7+8+9) = 24/45 2.4. 정확도 지표 2.4.1. 참 긍정 (True Positive) #TP = 7+8+9 = 24 2.4.2. 거짓 긍정 (False Positive) #FP = 1+2+3+4+5+6 = 21 2.4.3. 참 부정 (True Negative) #TN : 부정 예제에 대해 올바르게 다른 클래스라고 예상한 경우 2.4.4. 거짓 부정 (False Negative) #FN : 부정 예제에 대해 같은 클래스라고 잘못 예상한 경우 2.4.5. 정확도 (Accuracy) =(#TP + #TN) ÷ (#TP + #FP + #TN + #FN) 2.4.6. 민감도 (Sensitivity) = #TP ÷ (#TP + #FN) 2.4.7. 특이도 (Specificity) = #TN ÷ (#TN + #FP) 2.4.8. 정밀도 (Precision) = #TP ÷ (#TP + #FP) 2.4.9. 재현율 (Recall) = #TP ÷ (#TP + #FN) 2.4.10. F1 = 2 × 정밀도 × 재현율 ÷ (정밀도 + 재현율) = #TP ÷ [ #TP + (#FN + #FP)/2 ] 2.5. 수신자 조작 특성 곡선 2.6. 불균형 데이터세트 2.7. 정밀도, 진실도

class specific recall
https://www.researchgate.net/figure/Population-cumulative-confusion-matrix-showing-classification-accuracy-for-the-best_fig4_228812391

한글 좋다 http://bcho.tistory.com/1206

분류 결과표 Confusion Matrix
https://datascienceschool.net/view-notebook/731e0d2ef52c41c686ba53dcaf346f32/

http://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html#sphx-glr-auto-examples-model-selection-plot-confusion-matrix-py

https://www.quora.com/How-can-I-create-a-confusion-matrix-in-MS-Excel
Use =COUNTIFS. Example Predictions in column B and truths in column C. Create a 3x3 table in E1:G3. Truth positive value in E2, truth negative in E3. Prediction positive in F1 and prediction negative in G1. Top-left cell of the matrix (cell F2) would be =COUNTIFS ($B:$B,1,$C:$C,1) where 1 represents positive in a data point that is a true-positive. You maybe have TRUE or “Yes” or something else so change 1 to that on your formula. Bottom-left, =COUNTIFS ($B:$B,1,$C:$C,0). And I'll let you work out the rest!