def fit_transform(self, x): x = np.array(x) return em(x)
def test_impute_missing_values(self): """ After imputation, no NaN's should exist""" imputed = em(self.data_m) self.assertFalse(np.isnan(imputed).any())
# 调用impyute包中的数据补齐方法,主要有Multivariate Imputation,Expectation Maximization import numpy as np import time import copy from impyute.imputations.cs import mice from impyute.imputations.cs import em # 加载数据集 path = 'E:\\大论文相关\\LMVFM\\data\\compare'; x = np.loadtxt(path + '\\australian\\30%\\missing_other.txt'); x1 = copy.deepcopy(x) # Multivariate Imputation by Chained Equations time1 = time.clock(); data_mice = mice(x); time_mice = time.clock() - time1; #Expectation Maximization time1 = time.clock(); data_em = em(x1); time_em = time.clock() - time1; np.savetxt('data_em.txt',data_em,'%.2f'); np.savetxt('data_mice.txt',data_mice,'%.2f'); print(time_mice ,'----', time_em)
def test_return_type(self): """ Check return type, should return an np.ndarray""" imputed = em(self.data_m) self.assertTrue(isinstance(imputed, np.ndarray))
from impyute.datasets import random_uniform # from impyute.imputations.cs import random_imputation # from impyute.imputations.cs import mean_imputation from impyute.imputations.cs import em raw_data = random_uniform(shape=(5, 5), missingness="mcar", th=0.2) complete_data = em(raw_data)