【人工智能】机器学习之暴力调参案例

暴力调参案例

使用的数据集为

from sklearn.datasets import fetch_20newsgroups

因为在线下载慢，可以提前下载保存到

首先引入所需库

import numpy as np
import pandas as pd
defaultencoding = 'utf-8'
import matplotlib as mpl
import matplotlib.pyplot as plt
from sklearn.naive_bayes import MultinomialNB
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.model_selection import GridSearchCV
from sklearn.feature_selection import SelectKBest,chi2
import sklearn.metrics as metrics
from sklearn.datasets import fetch_20newsgroups
import sys

编码问题显示

if sys.getdefaultencoding() != defaultencoding:
    reload(sys)
    sys.setdefaultencoding(defaultencoding)
mpl.rcParams['font.sans-serif']=[u'simHei']
mpl.rcParams['axes.unicode_minus']=False

如果报错的话可以改为

import importlib,sys

if sys.getdefaultencoding() != defaultencoding:
    importlib.reload(sys)
    sys.setdefaultencoding(defaultencoding)
mpl.rcParams['font.sans-serif']=[u'simHei']
mpl.rcParams['axes.unicode_minus']=False

用来正常显示中文
mpl.rcParams[‘font.sans-serif’]=[u’simHei’]
用来正常正负号
mpl.rcParams[‘axes.unicode_minus’]=False

获取数据

#data_home="./datas/"下载的新闻的保存地址subset='train'表示从训练集获取新闻categories获取哪些种类的新闻
datas=fetch_20newsgroups(data_home="./datas/",subset='train',categories=['alt.atheism', 'comp.graphics', 'comp.os.ms-windows.misc'])
datas_test=fetch_20newsgroups(data_home="./datas/",subset='test',categories=['alt.atheism', 'comp.graphics', 'comp.os.ms-windows.misc'])
train_x=datas.data#获取新闻X
train_y=datas.target#获取新闻Y
test_x=datas_test.data#获取测试集的x
test_y=datas_test.target#获取测试集的y

自动调参

import time
def setParam(algo,name):
    gridSearch = GridSearchCV(algo,param_grid=[],cv=5)
    m=0
    if hasattr(algo,"alpha"):
        n=np.logspace(-2,9,10)
        gridSearch.set_params(param_grid={"alpha":n})
        m=10
    if hasattr(algo,"max_depth"):
        depth=[2,7,10,14,20,30]
        gridSearch.set_params(param_grid={"max_depth":depth})
        m=len(depth)
    if hasattr(algo,"n_neighbors"):
        neighbors=[2,7,10]
        gridSearch.set_params(param_grid={"n_neighbors":neighbors})
        m=len(neighbors)
    t1=time.time()
    gridSearch.fit(train_x,train_y)
    test_y_hat=gridSearch.predict(test_x)
    train_y_hat=gridSearch.predict(train_x)
    t2=time.time()-t1
    print(name, gridSearch.best_estimator_)
    train_error=1-metrics.accuracy_score(train_y,train_y_hat)
    test_error=1-metrics.accuracy_score(test_y,test_y_hat)
    return name,t2/5*m,train_error,test_error

选择算法调参

朴素贝叶斯，随机森林，KNN

algorithm=[("mnb",MultinomialNB()),("random",RandomForestClassifier()),("knn",KNeighborsClassifier())]
for name,algo in algorithm:
    result=setParam(algo,name)
    results.append(result)

可视化

#把名称，花费时间，训练错误率，测试错误率分别存到单个数组
names,times,train_err,test_err=[[x[i] for x in results] for i in  range(0,4)]

axes=plt.axes()
axes.bar(np.arange(len(names)),times,color="red",label="耗费时间",width=0.1)
axes.bar(np.arange(len(names))+0.1,train_err,color="green",label="训练集错误",width=0.1)
axes.bar(np.arange(len(names))+0.2,test_err,color="blue",label="测试集错误",width=0.1)
plt.xticks(np.arange(len(names)), names)
plt.legend()
plt.show()

代码整合：

#coding=UTF-8
import numpy as np
import pandas as pd
defaultencoding = 'utf-8'
import matplotlib as mpl
import matplotlib.pyplot as plt
from sklearn.naive_bayes import MultinomialNB
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.model_selection import GridSearchCV
from sklearn.feature_selection import SelectKBest,chi2
import sklearn.metrics as metrics
from sklearn.datasets import fetch_20newsgroups
import sys
import importlib,sys

if sys.getdefaultencoding() != defaultencoding:
    # reload(sys)
    importlib.reload(sys)
    sys.setdefaultencoding(defaultencoding)
mpl.rcParams['font.sans-serif']=[u'simHei']
mpl.rcParams['axes.unicode_minus']=False

#data_home="./datas/"下载的新闻的保存地址subset='train'表示从训练集获取新闻categories获取哪些种类的新闻
datas=fetch_20newsgroups(data_home="./datas/",subset='train',categories=['alt.atheism', 'comp.graphics', 'comp.os.ms-windows.misc'])
datas_test=fetch_20newsgroups(data_home="./datas/",subset='test',categories=['alt.atheism', 'comp.graphics', 'comp.os.ms-windows.misc'])
train_x=datas.data#获取新闻X
train_y=datas.target#获取新闻Y
test_x=datas_test.data#获取测试集的x
test_y=datas_test.target#获取测试集的y

tfidf=TfidfVectorizer(stop_words="english")
train_x=tfidf.fit_transform(train_x,train_y)#向量转化
test_x=tfidf.transform(test_x)#向量转化

print(train_x.shape)
best=SelectKBest(chi2,k=1000)#降维变成一千列

train_x = best.fit_transform(train_x,train_y)#转换
test_x = best.transform(test_x)

import time
def setParam(algo,name):
    gridSearch = GridSearchCV(algo,param_grid=[],cv=5)
    m=0
    if hasattr(algo,"alpha"):
        n=np.logspace(-2,9,10)
        gridSearch.set_params(param_grid={"alpha":n})
        m=10
    if hasattr(algo,"max_depth"):
        depth=[2,7,10,14,20,30]
        gridSearch.set_params(param_grid={"max_depth":depth})
        m=len(depth)
    if hasattr(algo,"n_neighbors"):
        neighbors=[2,7,10]
        gridSearch.set_params(param_grid={"n_neighbors":neighbors})
        m=len(neighbors)
    t1=time.time()
    gridSearch.fit(train_x,train_y)
    test_y_hat=gridSearch.predict(test_x)
    train_y_hat=gridSearch.predict(train_x)
    t2=time.time()-t1
    print(name, gridSearch.best_estimator_)
    train_error=1-metrics.accuracy_score(train_y,train_y_hat)
    test_error=1-metrics.accuracy_score(test_y,test_y_hat)
    return name,t2/5*m,train_error,test_error
results=[]
plt.figure()
algorithm=[("mnb",MultinomialNB()),("random",RandomForestClassifier()),("knn",KNeighborsClassifier())]
for name,algo in algorithm:
    result=setParam(algo,name)
    results.append(result)
#把名称，花费时间，训练错误率，测试错误率分别存到单个数组
names,times,train_err,test_err=[[x[i] for x in results] for i in  range(0,4)]

axes=plt.axes()
axes.bar(np.arange(len(names)),times,color="red",label="耗费时间",width=0.1)
axes.bar(np.arange(len(names))+0.1,train_err,color="green",label="训练集错误",width=0.1)
axes.bar(np.arange(len(names))+0.2,test_err,color="blue",label="测试集错误",width=0.1)
plt.xticks(np.arange(len(names)), names)
plt.legend()
plt.show()

结果：