import numpy as np import pandas as pd from scipy.stats import kstest #from sklearn import preprocessing # get a column from dataframe def select_data(data, ny): yName = data.columns[ny] Y = data[yName] return Y # see which feature is normally distributed from dataframe def normal_test(df): for i in range(len(df.columns)): y = select_data(df,i) p = kstest(y,'norm') print("feature {}, p-value = {}".format(i,p[1])) # rescale feature i in dataframe def standard_rescale(df, i): y = select_data(df,i) m = np.mean(y) s = np.std(y) y = (y-m)/s return y # log-transform feature of dataframe def log_transform(df,i): y = select_data(df,i) y = np.log(y) return y # square root transform feature of dataframe def sqrt_transform(df,i): y = select_data(df,i) y = np.sqrt(y) return y # cube root transform feature of dataframe def cbrt_transform(df,i): y = select_data(df,i) y = np.cbrt(y) return y # transform dataframe into one of: standard, log, sqrt, cbrt def transform_dataframe(df, transformation): df_new = [] if transformation == "standard": for i in range(len(df.columns)-1): y = standard_rescale(df,i) df_new.append(y) df_new.append(df.iloc[:,no_feats]) elif transformation == "log": for i in range(len(df.columns)-1): y = log_transform(df,i) df_new.append(y) df_new.append(df.iloc[:,no_feats]) elif transformation == "sqrt": for i in range(len(df.columns)-1): y = sqrt_transform(df,i) df_new.append(y) df_new.append(df.iloc[:,no_feats]) elif transformation == "cbrt": for i in range(len(df.columns)-1): y = cbrt_transform(df,i) df_new.append(y) df_new.append(df.iloc[:,no_feats]) else: return "wrong arguments" df_new = pd.DataFrame(df_new) df_new = df_new.T return df_new df = pd.read_csv('iris.csv') no_feats = 4 df.columns =['0', '1', '2', '3', '4'] #normal_test(df) df_standard = transform_dataframe(df, "standard") #df_log = transform_dataframe(df, "log") #df_sqrt = transform_dataframe(df, "sqrt") #df_cbrt = transform_dataframe(df, "cbrt") #df_wrong = transform_dataframe(df, "lo") #print("standard-----------------------------------------") #normal_test(df_standard) #print("log-----------------------------------------") #normal_test(df_log) #print("square root-----------------------------------------") #normal_test(df_sqrt) #print("cube root-----------------------------------------") #normal_test(df_cbrt) result = df_standard # create new csv file with new dataframe result.to_csv(r'iris_std.csv', index = False, header=True)解释每一行代码
最新发布
8042
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
