[人工智能]史上最全机器学习导包指南，新手看了都哭了。

Torch：?

from torch.utils.data import Dataset 数据读取模块

from torch.utils.data import DataLoader  数据分批

from torchvision import transforms
图像处理，灰度处理；缩放图像；90%的数据灰度化；将图像转换为tensor；归一化

import torch.nn as nn  导入网络模块

import torch.nn.functional as F  #F.relu/max_pool/fc

from torch.optim import SGD  # 随机梯度下降

图像处理：

from PIL import Image

import cv2

cv2.CascadeClassifier(" ")# 人脸检测
img = cv2.imread("girl.jpg")
cv2.imwrite("girl_v2.jpg", img)

# cv2.imshow("img", img)??# cv2.waitKey(0)

机器学习：

公用（公用部分）：

from sklearn.model_selection import train_test_split
#拆分数据集

from sklearn.preprocessing import StandardScaler 预处理

标准差标准化：

std = StandardScaler()
std.fit(X_train)
X_train = std.transform(X_train)
X_test = std.transform(X_test)

## 多项式特征：

from sklearn.preprocessing import PolynomialFeatures 

#CV网格搜索：

from sklearn.model_selection import GridSearchCV 
param_grid = {"criterion": ["gini", "entropy"],
"max_depth": range(2, 20)}
grid = GridSearchCV(dt, param_grid=param_grid,cv=10)
grid.fit(X_train, y_train)    print(grid.best_params_)

评估结果：

?

?

?

?

?

?PR曲线：

from sklearn.metrics import accuracy_score  # 准确率
from sklearn.metrics import recall_score  # 查全率
from sklearn.metrics import precision_score  # 查准率
from sklearn.metrics import confusion_matrix
from sklearn.metrics import classification_report  # 分类报告

from sklearn.metrics import plot_precision_recall_curve
plot_precision_recall_curve(alg, X_test, y_test)
plt.show() #画PR曲线

# 3. 观察相关性
print(data.corr()) # 相关性的计算方式：
{'pearson', 'kendall', 'spearman'}

from sklearn.metrics import mean_squared_error  # MSE
from sklearn.metrics import mean_absolute_error  # MAE
print("均方误差", mean_squared_error(y_test, y_pred))
print("均方根误差RMSE", np.sqrt(mean_squared_error(y_test, y_pred)))
print("平均绝对误差MAE", mean_absolute_error(y_test, y_pred))
print("R2", lr.score(X_test,y_test))

算法类的：

Xgbboost:

from xgboost import XGBClassifier
from xgboost import XGBRegressor
from xgboost import plot_tree
from xgboost import plot_importance #画树图
import matplotlib.pyplot as plt
from xgboost import to_graphviz 

randomFrest:

?

?

?

from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import RandomForestRegressor

dot_file = to_graphviz(xgb, num_trees=1, fmap='iris_xgb_v2.fmap')
dot_file.render("iris_xgb.dot")

决策树：

from sklearn.tree import DecisionTreeClassifier
from sklearn.tree import export_graphviz

?词频向量化：

import jieba  #中文分词
from sklearn.feature_extraction.text import CountVectorizer  # 词频向量化
import joblib

out = " ".join(list(jieba.cut(text))) 
cnt = CountVectorizer(stop_words=stop_words)
cnt.fit(train_X)
train_X = cnt.transform(train_X).toarray()
test_X = cnt.transform(test_X).toarray()

joblib.dump(cnt, "词频向量化.model") # 保存词频向量化
joblib.dump(alg, "垃圾邮件分类.model") # 保存模型

jcnt = joblib.load("词频向量化.model")
test_text = cnt.transform(test_text).toarray()

alg = joblib.load("垃圾邮件分类.model")
print("预测结果", alg.predict(test_text))

贝叶斯：

from sklearn.naive_bayes import MultinomialNB #多项式

?

?

?

?

?逻辑回归：

from sklearn.linear_model import LogisticRegression  
# 逻辑回归
from sklearn.linear_model import SGDRegressor  
# 随机梯度下降的线性回归
from sklearn.linear_model import LinearRegression 
 # 正规方程的线性回归

回归算法的评估导包：

?

from sklearn.linear_model import Ridge  # 岭回归  
 基础回归+L2正则化
from sklearn.linear_model import Lasso  # 套索回归  
 基础回归+L1正则化
# pf = PolynomialFeatures(degree=2)
# X = pf.fit_transform(X) #特征多项式增维度

KNN算法：

from sklearn.neighbors import KNeighborsClassifier  # KNN的分类算法

?

?k-means ：

from sklearn.cluster import KMeans

?

?

from sklearn.metrics import silhouette_score
silhouette_score(train, y_pred) # 轮廓系数 范围【-1,1】
； 最好情况是1

#TD-idf ：

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.feature_extraction.text import TfidfTransformer

词云 ：

from wordcloud import WordCloud

?

# 从文本中生成词云文本
wc_text = wc.generate_from_text(texts)
plt.imshow(wc_text) 
plt.show()

#pyecharts 词云 ：

from pyecharts.charts import WordCloud
cnt = Counter(texts)  # 每个词出现的次数
wc = WordCloud()
wc.add(series_name="",
       data_pair=out,
       mask_image="../ML_data/qie.png",
       )
wc.render("中文词云.html")

?

?