查看“模型评估指标”的源代码

混淆矩阵(Confusion matrix, error matrix)是一种具有两个维度的(实际与预测)列联表，矩阵的每一列代表一个类的实例预测，而每一行表示一个实际的类的实例。通过这个矩阵可以方便地看出模型是否将两个不同的类混淆了。
{| class="wikitable"
|+
!
!
! colspan="2" |Predicted condition
|-
|
|'''Total population = P + N'''
|Positive (PP)
|Negative (PN)
|-
| rowspan="2" |'''Actual condition'''
|Positive (P)
|True positive (TP), 真阳性
|False negative (FN), 假阴性
|-
|Negative (N)
|False positive (FP), 假阳性
|
|}

==== 精准率 ====
精准率(precision)：预测为阳性的数据中预测正确的数据个数

招回率(recall)：阳性的数据中预测正确的数据个数
====Sample====
{| class="wikitable" style="text-align:center;"
!Individual Number
!1
!2
!3
!4
!5
!6
!7
!8
!9
!10
!11
!12
|-
! style="background:#eeeebb" |Actual Classification
| style="background:#ffffcc" |1
| style="background:#ffffcc" |1
| style="background:#ffffcc" |1
| style="background:#ffffcc" |1
| style="background:#ffffcc" |1
| style="background:#ffffcc" |1
| style="background:#ffffcc" |1
| style="background:#ffffcc" |1
| style="background:#ddddaa" |0
| style="background:#ddddaa" |0
| style="background:#ddddaa" |0
| style="background:#ddddaa" |0
|-
! style="background:#bbeeee" |Predicted Classification
| style="background:#aadddd" |0
| style="background:#aadddd" |0
| style="background:#ccffff" |1
| style="background:#ccffff" |1
| style="background:#ccffff" |1
| style="background:#ccffff" |1
| style="background:#ccffff" |1
| style="background:#ccffff" |1
| style="background:#ccffff" |1
| style="background:#aadddd" |0
| style="background:#aadddd" |0
| style="background:#aadddd" |0
|}
{| class="wikitable"
!
!
! colspan="2" |Predicted condition
|-
|
|'''Total population = 8 + 4'''
|Positive (PP)
|Negative (PN)
|-
| rowspan="2" |'''Actual condition'''
|Cancer 8
|6
|2
|-
|Non-cancer 4
|1
|3
|}

[[分类:Develop]]
[[分类:Algorithm]]