数据科学包整理

##导入数据
from sklearn.datasets import 内置数据集

##分类
#朴素贝叶斯
from sklearn.naive_bayes import GaussianNB#高斯朴素贝叶斯
from sklearn.naive_bayes import MultinomialNB#多项式朴素贝叶斯
#决策树
from sklearn.tree import DecisionTreeClassifier#决策树分类
from sklearn.tree import export_graphviz#输出模型
#逻辑回归
from sklearn.linear_model import LogisticRegression     
#支持向量机
from sklearn.svm import SVC#支持向量分类器
#神经网络
from sklearn.neural_network import MLPClassifier
#集成算法-随机森林
from sklearn.ensemble import RandomForestClassifier
#集成算法-AdaBoost分类
from sklearn.ensemble import AdaBoostClassifier
    
##异常检测
from sklearn.svm import OneClassSVM #一类支持向量机
    
    
##回归
#[线型回归](https://blog.csdn.net/weixin_40844116/article/details/108714628)
from sklearn.linear_model import LinearRegression 
##回归模型评估        
import statsmodels.api as sm 
    
#岭回归    
from sklearn.linear_model import Ridge 
#贝叶斯岭回归
from sklearn.linear_model import BayesianRidge
#弹性网络回归
from sklearn.linear_model import ElasticNet
#支持向量机回归    
from sklearn.svm import SVR
#集成-梯度提升回归    
from sklearn.ensemble.gradient_boosting import GradientBoostingRegressor
  
    
    
##聚类
#层次聚类
import scipy.cluster.hierarchy as sch    
#矢量聚类
#import scipy.cluster.vq 
#层次聚类-凝聚
from sklearn.cluster import AgglomerativeClustering
#kmeans聚类
from sklearn.cluster import KMeans
#DBSCAN
from sklearn.cluster import DBSCAN
    

##模型评估
#准确得分
from sklearn.metrics import accuracy_score
#混淆矩阵
from sklearn.metrics import confusion_matrix
#准确率和召回率（precision_score\recall_score\f1_score）
from sklearn.metrics import classification_report    
#roc曲线
from sklearn.metrics import roc_curve 
#auc分数
from sklearn.metrics import auc    
#余弦相似度
from sklearn.metrics.pairwise import cosine_similarity
#解释方差得分：[0,1]值越小效果越差    
from sklearn.metrics import explained_variance_score
#平均绝对误差：预测数和真实数接近程度，值越小拟合效果越好
from sklearn.metrics import mean_absolute_error
#均方误差：拟合数和真实数对应样本点误差的平方和的均值，值越小拟合效果越好
from sklearn.metrics import mean_squared_error
#判定系数得分：[0,1]值越小效果越差   
from sklearn.metrics import r2_score

##模型数据选择
#训练集-测试集拆分
from sklearn.model_selection import train_test_split
#KFold拆分
from sklearn.model_selection import KFold
#分层KFold拆分-应对分类样本不均衡    
from sklearn.model_selection import StratifiedKFold
#交叉验证得分
from sklearn.model_selection import cross_val_score
#留一拆分
from sklearn.model_selection import LeaveOneOut

##特征选择
#方差选择法-移除低变异特征
from sklearn.feature_selection import VarianceThreshold
#K最佳选择法
from sklearn.feature_selection import SelectKBest   
#卡方统计量
from sklearn.feature_selection import chi2
##特征抽取
#分类特征
from sklearn.feature_extraction import DictVectorizer
#文本特征-词频矩阵
from sklearn.feature_extraction.text import CountVectorizer    
#根据最高分数的百分位数选择特征
from sklearn.feature_selection import SelectPercentile , f_classif#f_classif计算数据集的方差P值
##降维
#主成分分析PCA
from sklearn.decomposition import PCA
#奇异值
from scipy.linalg import svd#SVD分解
from sklearn.decomposition import TruncatedSVD#SVD降维


#关联规则
from apyori import apriori
#频繁样式探勘
from pymining import itemmining


##[管道](https://blog.csdn.net/weixin_40844116/article/details/108362986)
from sklearn.pipeline import make_pipeline
    
    
##预处理    
#[多项式转换器](https://blog.csdn.net/weixin_40844116/article/details/108714628)
from sklearn.preprocessing import PolynomialFeatures
#标准化
from sklearn.preprocessing import StandardScaler
#标签编码（转换成0开始的数字编码）
from sklearn.preprocessing import LabelEncoder
#二值化（0-1）
from sklearn.preprocessing import binarize
#独热编码
from sklearn.preprocessing import OneHotEncoder#将分类变量和顺序变量转换为二值化标志变量
#填充缺失值
from sklearn.preprocessing import Imputer