zhangxiaowbl / roc_curve_with_confidence_intervals

roc_curve_with_confidence_intervals

Python version

import platform
print(platform.python_version())

3.7.5

Notes

Run you jupyter notebook positioned on the stackoverflow project folder.

The the following notebook cell will append to your path the current folder where the jupyter notebook is runnig, in order to be able to import auc_delong_xu.py script for this example.

import os
import sys
import pandas as pd
import numpy as np
from sklearn import datasets

notebook_folder_path = !pwd
prj_path = os.path.abspath(os.path.join(notebook_folder_path[0], '', ''))
sys.path.append(prj_path)
print('Append to path: %s' % prj_path)

Append to path: /Users/lsanchez/roc_curve_with_confidence_intervals

The data

I used the iris dataset to create a binary classification task where the possitive class corresponds to the setosa class.

The y_score is simply the sepal length feature rescaled between [0, 1].

data = pd.DataFrame(
    datasets.load_iris().data,
    columns=datasets.load_iris().feature_names)
target = pd.Series([
    datasets.load_iris().target_names[x]
    for x in datasets.load_iris().target])

data.head()

	sepal length (cm)	sepal width (cm)	petal length (cm)	petal width (cm)
0	5.1	3.5	1.4	0.2
1	4.9	3.0	1.4	0.2
2	4.7	3.2	1.3	0.2
3	4.6	3.1	1.5	0.2
4	5.0	3.6	1.4	0.2

The AUC and Delong Confidence Interval is calculated via the Yantex's implementation of Delong (see script: auc_delong_xu.py for further details)

from auc_delong_xu import auc_ci_Delong

y_true = (target != 'setosa').astype(int)

y_score = data['sepal length (cm)']
y_score = (y_score - y_score.min()) / (y_score.max() - y_score.min())

y_pred = (y_score > .5).astype(int)

auc, auc_var, ci = auc_ci_Delong(
    y_true=y_true,
    y_scores=y_score)

print('ROC AUC: %s, Conf.' % auc) 
print('Confidence Interval: %s (95%% confidence)' % str(ci))

ROC AUC: 0.9586000000000001, Conf.
Confidence Interval: [0.93020851 0.98699149] (95% confidence)

R Version

version

               _                           
platform       x86_64-apple-darwin13.4.0   
arch           x86_64                      
os             darwin13.4.0                
system         x86_64, darwin13.4.0        
status                                     
major          3                           
minor          6.1                         
year           2019                        
month          07                          
day            05                          
svn rev        76782                       
language       R                           
version.string R version 3.6.1 (2019-07-05)
nickname       Action of the Toes          

The data

I used the iris dataset to create a binary classification task where the possitive class corresponds to the setosa class.

The y_score is simply the sepal length feature rescaled between [0, 1].

library(pROC)
library(datasets)

data(iris)

y_true = as.integer(iris$Species == 'setosa')
y_score = iris$Sepal.Length
y_score = (y_score - min(y_score)) / (max(y_score) - min(y_score))

y_pred = as.integer(y_score > .5)

roc = roc(y_true, y_score)
roc

Type 'citation("pROC")' for a citation.

Attaching package: ‘pROC’

The following objects are masked from ‘package:stats’:

    cov, smooth, var

Setting levels: control = 0, case = 1
Setting direction: controls > cases




Call:
roc.default(response = y_true, predictor = y_score)

Data: y_score in 100 controls (y_true 0) > 50 cases (y_true 1).
Area under the curve: 0.9586

print(ci(roc))

95% CI: 0.9302-0.987 (DeLong)