论文复现：Ensemble-based active learning using fuzzy-rough approach for cancer sample classification

 Reference:Ansuman Kumar, Anindya Halder.Ensemble-based active learning using fuzzy-rough approach for cancer sample classification.Engineering Applications of Artificial Intelligence, 2020 (91) 103591

不用查了，是SCI二区文章。

不用读了。作者写功好而已

文章在参数设置方面讲的不清楚。

1、模糊数m设置没有讲。

2、每个子模型的k值设置为多少没有讲。

3、集成过程并集为空怎么办没有讲。

花一小时复现了个非集成了的。

读者感兴趣的，自己改成集成的即可（用不同的k值重复select()函数）

'''
author:Deniu He
date:2020-12-11
organization: CQUPT
Reference:Ansuman Kumar, Anindya Halder.Ensemble-based active learning using fuzzy-rough approach for cancer
sample classification.Engineering Applications of Artificial Intelligence, 2020 (91) 103591
'''
import xlwt
import numpy as np
import pandas as pd
from copy import deepcopy
from sklearn.metrics import accuracy_score, mean_absolute_error
from collections import OrderedDict
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import StratifiedKFold
from pathlib import Path
from scipy.spatial.distance import  pdist,squareform


class ALFRNN():
    def __init__(self,X_pool,y_pool,labeled,budget,X_test,y_test):
        self.X_pool = X_pool
        self.y_pool = y_pool.astype(np.int)
        self.X_test = X_test
        self.y_test = y_test.astype(np.int)
        self.labels = np.unique(y_pool)
        self.n_class = len(np.unique(y_pool))
        self.labeled = list(deepcopy(labeled))
        self.unlabeled = self.initialization()
        self.distMatrix = squareform(pdist(self.X_pool,metric='euclidean'))
        self.budgetLeft = deepcopy(budget)
        self.budget = deepcopy(budget)
        self.model = LogisticRegression()
        self.AccList = []
        self.MAEList = []

        self.ALC_Acc = 0.0
        self.ALC_MAE = 0.0
        self.ALC_Acc_10k = 0.0
        self.ALC_MAE_10k = 0.0

    def initialization(self):
        unlabeled = [i for i in range(len(self.y_pool))]
        for j in self.labeled:
            unlabeled.remove(j)
        return unlabeled

    def find_neighbor(self,k):
        neighbor = OrderedDict()
        for idx in self.unlabeled:
            neighbor[idx] = []
            dist_list = []
            for j, jdx in enumerate(self.labeled):
                dist_list.append(self.distMatrix[idx,jdx])
            ordjdx = np.argsort(dist_list)
            for r in range(k):
                neighbor[idx].append(self.labeled[ordjdx[r]])
        return neighbor

    def C(self,j,jdx):
        if self.y_pool[jdx] == self.labels[j]:
            return 1.0
        else:
            return 0.0

    def select(self):
        while self.budgetLeft > 0:
            neighbor = self.find_neighbor(k=self.n_class-1)
            R = OrderedDict()
            for idx, nei in neighbor.items():
                dist_list = []
                for jdx in nei:
                    dist_list.append(self.distMatrix[idx,jdx])
                dist_list = np.array(dist_list)
                dist_list = dist_list ** 2
                R[idx] = sum(dist_list)/dist_list

            lower = OrderedDict()
            for idx, nei in neighbor.items():
                lower[idx] = np.zeros(self.n_class)
                for j in range(self.n_class):
                    rc = []
                    for b,jdx in enumerate(nei):
                        rc.append(min([1,1-R[idx][b]+self.C(j=j,jdx=jdx)]))
                    lower[idx][j] = min(rc)

            upper = OrderedDict()
            for idx, nei in neighbor.items():
                upper[idx] = np.zeros(self.n_class)
                for j in range(self.n_class):
                    rc = []
                    for b, jdx in enumerate(nei):
                        rc.append(max([0,R[idx][b]+self.C(j=j,jdx=jdx)-1]))
                    upper[idx][j] = max(rc)

            Avg = OrderedDict()
            metric = OrderedDict()
            for idx in self.unlabeled:
                Avg[idx] = np.zeros(self.n_class)
                for j in range(self.n_class):
                    Avg[idx][j] = (lower[idx][j] + upper[idx][j])/2
                Avg[idx] = Avg[idx]/ max(Avg[idx])
                ordjx = np.argsort(Avg[idx])
                metric[idx] = Avg[idx][ordjx[1]]
            tar_idx = max(metric,key=metric.get)
            self.labeled.append(tar_idx)
            self.unlabeled.remove(tar_idx)
            self.budgetLeft -= 1
            # print("预算剩余：{}".format(self.budgetLeft))
            self.model.fit(X=self.X_pool[self.labeled], y=self.y_pool[self.labeled])
            Acc = accuracy_score(y_true=self.y_test, y_pred=self.model.predict(self.X_test))
            MAE = mean_absolute_error(y_true=self.y_test, y_pred=self.model.predict(self.X_test))
            self.AccList.append(Acc)
            self.MAEList.append(MAE)
            self.ALC_Acc += Acc
            self.ALC_MAE += MAE