import pandas as pd
import numpy as np
from sklearn.preprocessing import MinMaxScaler
data = pd.DataFrame(np.random.randint(10, 100, (5, 2)))  # 準備測試數(shù)據(jù)
# 常用接口
scaler = MinMaxScaler()  # 默認范圍為 [0,1]
scaler = MinMaxScaler(feature_range=[5, 10])  # 自定義歸一化數(shù)據(jù)范圍
scaler.fit(data)  # 擬合數(shù)據(jù)
scaler.partial_fit(data)  # 數(shù)據(jù)量大的時候fit()報錯，可以使用partial_fit()
result = scaler.transform(data)   # 變換數(shù)據(jù)
result = scaler.fit_transform(data)    # 擬合和變換一步達成
data = scaler.inverse_transform(result)  # 逆向變換

import pandas as pd
import numpy as np
from sklearn.preprocessing import StandardScaler
data = pd.DataFrame(np.random.randint(10, 100, (5, 2)))  # 準備測試數(shù)據(jù)
# 常用接口
scaler = StandardScaler()  # 創(chuàng)建對象
scaler.fit(data)  # 擬合數(shù)據(jù)
scaler.partial_fit(data)  # 數(shù)據(jù)量大的時候fit()報錯，可以使用partial_fit()
result = scaler.transform(data)  # 變換數(shù)據(jù)
result = scaler.fit_transform(data)  # 擬合和變換一步達成
data = scaler.inverse_transform(result)  # 逆向變換
# 常用屬性
scaler.var_  # 擬合后查看各列數(shù)據(jù)的方差
scaler.mean_  # 擬合后查看各列數(shù)據(jù)的均值

import pandas as pd
import numpy as np
from sklearn.impute import SimpleImputer
data = pd.DataFrame(np.random.randint(10, 100, (5, 2)))  # 準備測試數(shù)據(jù)
# 常用接口
imp_mean = SimpleImputer()  # 均值填充缺失值
imp_median = SimpleImputer(strategy='median')  # 中位數(shù)填充缺失值
imp_mode = SimpleImputer(strategy='most_frequent')  # 眾數(shù)填充
imp_0 = SimpleImputer(strategy='constant', fill_value=0)  # 0填充
imp_mean.fit(data)  # 擬合數(shù)據(jù)
result = imp_mean.transform(data)  # 變換數(shù)據(jù)
result = imp_mean.fit_transform(data)  # 擬合和變換一步到位

import pandas as pd
import numpy as np
data = pd.DataFrame(np.random.randint(10, 100, (5, 2)))  # 準備測試數(shù)據(jù)
data.iloc[0:2, 0:2] = np.nan
# 使用pandas進行缺失值填充
result = data.fillna(0)  # 0填充所有缺失值
# 均值填充第0列,中位數(shù)填充、眾數(shù)填充同理
result = data.iloc[:, 0] = data.iloc[:, 0].fillna(data.iloc[:, 0].mean())  
result = data.dropna(axis=0)  # 刪除所有含有缺失值的樣本數(shù)據(jù)

import pandas as pd
from sklearn.preprocessing import LabelEncoder, OrdinalEncoder, OneHotEncoder
# 準備測試數(shù)據(jù)
data_dic = {
    '性別': ['男', '女', '女', '男', '女'],
    '學歷': ['小學', '初中', '初中', '大學', '高中'],
    '標簽': ['標簽3', '標簽1', '標簽2', '標簽3', '標簽2'],
}
data = pd.DataFrame(data_dic) 
# 常用接口
le = LabelEncoder()  # 創(chuàng)建對象
oe = OrdinalEncoder()
le = le.fit(data.loc[:, '標簽'])  # 擬合數(shù)據(jù)
result = le.transform(data.loc[:, '標簽'])  # 變換數(shù)據(jù)
result = le.fit_transform(data.loc[:, '標簽'])  # 擬合和變換一步達成
data = le.inverse_transform(result)  # 逆向變換
data = pd.DataFrame(data_dic)
# 一般這樣寫
result = LabelEncoder().fit_transform(data.loc[:, '標簽']) 
result = OrdinalEncoder().fit_transform(data.loc[:, ['性別', '學歷']])
# 常用屬性
le.classes_   # 返回原來的字符串標簽唯一值數(shù)組, 按該數(shù)組順序編號
oe.categories_    # 返回原來的字符串標簽數(shù)組, 按該數(shù)組順序編號

import pandas as pd
from sklearn.preprocessing import OneHotEncoder
# 準備測試數(shù)據(jù)
data_dic = {
    '性別': ['男', '女', '女', '男', '女'],
    '學歷': ['小學', '初中', '初中', '大學', '高中'],
    '標簽': ['標簽3', '標簽1', '標簽2', '標簽3', '標簽2'],
}
data = pd.DataFrame(data_dic) 
# 常用接口
encoder = OneHotEncoder()
encoder.fit(data[['性別']])  # 擬合數(shù)據(jù)
result = encoder.transform(data[['性別']])  # 變換數(shù)據(jù)
result = encoder.fit_transform(data[['性別']])  # 擬合和變換一步到位
data = encoder.inverse_transform(result)  # 逆向變換
# 一般這樣寫
result = OneHotEncoder().fit_transform(data[['性別']])

import numpy as np
from sklearn.preprocessing import Binarizer
data = np.arange(10, 20).reshape(-1, 1)
bin = Binarizer(threshold=15)  # 默認threshold為0
bin.fit(data)  # 擬合數(shù)據(jù)
result = bin.transform(data)  # 變換數(shù)據(jù)
result = bin.fit_transform(data)  # 擬合和變換一步達成
# 一般這樣用
result = Binarizer(threshold=15).fit_transform(data)

import numpy as np
import pandas as pd
from sklearn.preprocessing import KBinsDiscretizer
data = pd.DataFrame(np.random.randint(10, 100, (5, 2)))  # 準備測試數(shù)據(jù)
est = KBinsDiscretizer(n_bins=3, encode='ordinal', strategy='quantile')
est.fit(data)  # 擬合數(shù)據(jù)
result = est.transform(data)  # 變換數(shù)據(jù)
result = est.fit_transform(data)  # 擬合和變換一步到位
data = est.inverse_transform(result)  # 
# 一般這樣用
result = KBinsDiscretizer(n_bins=3, encode='ordinal', strategy='quantile').fit_transform(data)