import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

# 讀取CSV文件
data = pd.read_csv('data.csv')

# 查看數(shù)據(jù)的前幾行
print(data.head())

# 查看數(shù)據(jù)的基本信息
print(data.info())

# 查看數(shù)據(jù)的統(tǒng)計(jì)描述
print(data.describe())

# 查看缺失值情況
print(data.isnull().sum())

# 刪除含有缺失值的行
data_cleaned = data.dropna()

# 用均值填充缺失值
data_filled = data.fillna(data.mean())

# 使用插值法填充缺失值
data_interpolated = data.interpolate()

# 刪除重復(fù)值
data_deduplicated = data.drop_duplicates()

# 將字符串類型的日期轉(zhuǎn)換為日期類型
data['date'] = pd.to_datetime(data['date'])

# 將分類數(shù)據(jù)轉(zhuǎn)換為數(shù)值類型
data['category'] = data['category'].astype('category').cat.codes

from sklearn.preprocessing import StandardScaler, MinMaxScaler

# 標(biāo)準(zhǔn)化
scaler = StandardScaler()
data_standardized = scaler.fit_transform(data)

# 歸一化
scaler = MinMaxScaler()
data_normalized = scaler.fit_transform(data)

# 使用箱線圖法檢測(cè)和處理異常值
Q1 = data.quantile(0.25)
Q3 = data.quantile(0.75)
IQR = Q3 - Q1
data_outlier_removed = data[~((data < (Q1 - 1.5 * IQR)) |(data > (Q3 + 1.5 * IQR))).any(axis=1)]

# 使用Z分?jǐn)?shù)法檢測(cè)和處理異常值
from scipy import stats
data_zscore = data[(np.abs(stats.zscore(data)) < 3).all(axis=1)]

# 創(chuàng)建新特征：日期特征分解
data['year'] = data['date'].dt.year
data['month'] = data['date'].dt.month
data['day'] = data['date'].dt.day

# 特征組合
data['total_amount'] = data['quantity'] * data['price']

# 繪制直方圖
data['column_name'].hist()
plt.show()

# 繪制箱線圖
data.boxplot(column='column_name')
plt.show()

# 繪制散點(diǎn)圖
plt.scatter(data['column1'], data['column2'])
plt.show()

# 計(jì)算與目標(biāo)變量的相關(guān)系數(shù)
correlation = data.corr()
print(correlation['target_variable'].sort_values(ascending=False))

from sklearn.linear_model import Lasso

# 使用Lasso進(jìn)行特征選擇
lasso = Lasso(alpha=0.1)
lasso.fit(data.drop('target_variable', axis=1), data['target_variable'])
selected_features = data.columns[lasso.coef_ != 0]
print(selected_features)

from sklearn.feature_selection import RFE
from sklearn.linear_model import LogisticRegression

# 使用RFE進(jìn)行特征選擇
model = LogisticRegression()
rfe = RFE(model, 5)
fit = rfe.fit(data.drop('target_variable', axis=1), data['target_variable'])
selected_features = data.columns[fit.support_]
print(selected_features)

from sklearn.model_selection import train_test_split

# 數(shù)據(jù)分割
X_train, X_test, y_train, y_test = train_test_split(data.drop('target_variable', axis=1), data['target_variable'], test_size=0.2, random_state=42)

import pandas as pd
import numpy as np
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split

# 讀取數(shù)據(jù)
data = pd.read_csv('data.csv')

# 數(shù)據(jù)探索
print(data.info())
print(data.describe())

# 處理缺失值
data = data.fillna(data.mean())

# 刪除重復(fù)值
data = data.drop_duplicates()

# 數(shù)據(jù)類型轉(zhuǎn)換
data['date'] = pd.to_datetime(data['date'])
data['category'] = data['category'].astype('category').cat.codes

# 特征工程
data['year'] = data['date'].dt.year
data['month'] = data['date'].dt.month
data['day'] = data['date'].dt.day
data['total_amount'] = data['quantity'] * data['price']

# 處理異常值
Q1 = data.quantile(0.25)
Q3 = data.quantile(0.75)
IQR = Q3 - Q1
data = data[~((data < (Q1 - 1.5 * IQR)) | (data > (Q3 + 1.5 * IQR))).any(axis=1)]

# 數(shù)據(jù)標(biāo)準(zhǔn)化
scaler = StandardScaler()
data_scaled = scaler.fit_transform(data.drop(['date', 'target_variable'], axis=1))

# 數(shù)據(jù)分割
X_train, X_test, y_train, y_test = train_test_split(data_scaled, data['target_variable'], test_size=0.2, random_state=42)