Python实现AHP（层次分析法）
16lz
2021-01-22
import csv import numpy as np import tensorflow as tf from sklearn.preprocessing import MinMaxScaler from numpy import * class AHP: def __init__(self, array): self.row = len(array) self.col = len(array[0]) def get_tezheng(self, array):# 获取特征值和特征向量 te_val, te_vector = np.linalg.eig(array) list1 = list(te_val) print("特征值为：", te_val) print("特征向量为：", te_vector) # 得到最大特征值对应的特征向量 max_val = np.max(list1) index = list1.index(max_val) max_vector = te_vector[:, index] print("最大的特征值:" + str(max_val) + " 对应的特征向量为：" + str(max_vector)) return max_val, max_vector def RImatrix(self, n): # 建立RI矩阵 print(n) n1 = [1, 2, 3, 4, 5, 6, 7, 8, 9] n2 = [0, 0, 0.58, 0.90, 1.12, 1.24, 1.32, 1.41, 1.45] d = dict(zip(n1,n2)) print("该矩阵在一致性检测时采用的RI值为：", d[n]) return d[n] def test_consitstence(self, max_val, RI): # 测试一致性 CI = (max_val - self.row) / (self.row - 1) if RI == 0: print("判断矩阵的RI值为 " + str(0) + " 通过一致性检验") return True else: CR = CI / RI if CR < 0.10: print("判断矩阵的CR值为 " + str(CR) + " 通过一致性检验") return True else: print("判断矩阵的CR值为 " + str(CR) + " 判断矩阵未通过一致性检验，请重新输入判断矩阵") return False def main(): weigh_matrix = [] num_of_standard = [] l0 = [] l1 = [] def input_data(f1): # 对输入的判断矩阵进行处理 l0 = [] l1 = [] li = [] with open(f1) as f: reader = csv.reader(f) for line in reader: print(line) if line == []: l1.append(l0) l0 = [] continue for i in line: if "/" in i: b = i.split("/") li.append(int(b[0]) / float(b[1])) else: li.append(float(i)) l0.append(li) li = [] l1.append(l0) print(l1) return l1 l1= input_data(r"C:\Users\adm\Desktop\data11.txt") print("l1",l1) def normalize_vector(max_vector): # 特征向量归一化 vector_after_normalization = [] sum0 = np.sum(max_vector) for i in range(len(max_vector)): vector_after_normalization.append(max_vector[i] / sum0) print("该级指标的权重权重矩阵为： " + str(vector_after_normalization)) file = open(r"C:\Users\adm\Desktop\data000112.txt", 'a') for i in vector_after_normalization: file.write(str(i)) file.write("\n") return vector_after_normalization def to_input_matrix():#判别矩阵的一致性检验 for i in l1: length = len(l1) print(i) a = AHP(i) max_val, max_vector = a.get_tezheng(i) record_max_vector = max_vector RI = a.RImatrix(len(i)) flag = a.test_consitstence(max_val, RI) while not flag: print("对比矩阵未通过一致性检验，请重新输入对比矩阵！") break #flag = to_input_matrix(length) weight = normalize_vector(record_max_vector) # 返回权重[[0.5, 0.5], [0.5, 0.5], [0.5, 0.5]] weigh_matrix.append(weight) print("最终的权重矩阵为：", weigh_matrix) return weigh_matrix w00=to_input_matrix() print("w00",w00) def construct_weightmatrix(weight): data = [] l = [] for i in weight[0]: data.append(i.real) #print(data) for i in weight[1:4]: for j in i: data.append(j.real) for i in weight[4:]: for j in i: data.append(j.real) print(data) data2 = data print(len(data2)) return data2 d = construct_weightmatrix(w00) print(d) def get_index(data ): # 第一个矩阵1*3，第二个矩阵7*3。第三个矩阵16*7 row = [3, 7, 16] column = [1, 3, 7] x1 = [0, 1, 2] y1 = [0, 0, 0] x2 = [0, 1, 2, 3, 4, 5, 6] y2 = [0, 0, 1, 1, 1, 2, 2] x3 = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, ] y3 = [0, 0, 1, 1, 2, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 6] weight = [] def transform(x, y, data, row, col): h = list(zip(x, y)) print(h) mat1 = zeros((row, col)) for i in range(len(data)): mat1[h[i]] = data[i] weight.append(mat1) print(weight) transform(x1, y1, data[:3], 3, 1) transform(x2, y2, data[3:10], 7, 3) transform(x3, y3, data[10:], 16, 7) return weight w1 = get_index(d)#返回组织好的权重矩阵d for i in w1: print(i) # print(w1) def get_finalweight(w):#矩阵相乘 j = 0 s = np.array([1]) print(np.dot(w[1], w[0])) while j < len(w): print("weight0[i]", w[j]) s = np.dot(w[j], s) print("s", s) j += 1 print("s", s) return s s1=get_finalweight(w1) def data_process(): i = 0 data = [] global name name = [] with open(r'C:\Users\adm\Desktop\数据.csv') as f: reader = csv.reader(f) for row in reader: if i == 0: i += 1 continue else: data.append(row[:]) name.append(row[1]) data = np.asarray(data) data1 = data[:, 3:].astype('float64') scaler = MinMaxScaler(feature_range=(0, 1)) data1 = scaler.fit_transform(data1) print(len(data1)) return data1 def ScoreCalculator(): SCORE = [] score = [] value = data_process() print(len(value)) for i in range(len(value)): #print("i",len(value[i])) #print(len(s1)) SCORE = np.sum([x * y for x, y in zip(value[i], s1)]) * 100 score.append(SCORE) score = [i / np.max(score) * 100 for i in score] print(score) d = dict(zip(name, score)) d1 = sorted(d.items(), key=lambda items: items[1], reverse=True) # print(d1) for key, value in d1[:80]: if value > 80: print(str(key) + " 最终得到的评分为： " + str(value) + " 合理") elif 70 < value < 80: print(str(key) + " 最终得到的评分为： " + str(value) + " 较为合理") elif 60 < value < 70: print(str(key) + " 最终得到的评分为： " + str(value) + " 一般合理") else: print(str(key) + " 最终得到的评分为： " + str(value) + " 不合理") print("所有指标权重的和", np.sum(s1)) ScoreCalculator() if __name__ == "__main__": main()