numpy 矩阵归一化

new_value = (value - min)/(max-min)

 

def normalization(datingDatamat):
   max_arr = datingDatamat.max(axis=0)
   min_arr = datingDatamat.min(axis=0)
   ranges = max_arr - min_arr
   norDataSet = zeros(datingDatamat.shape)
   m = datingDatamat.shape[0]
   norDataSet = datingDatamat - np.tile(min_arr, (m, 1))
   norDataSet = norDataSet/np.tile(ranges,(m,1))
   return norDataSet

转载于:https://www.cnblogs.com/earendil/p/8311635.html