用泰坦尼克号事件的数据集练习一下决策树和随机森林的API。
分析数据集信息
先读入数据集，看看有哪些特征：
import pandas as pd
titanic = pd.read_csv("titanic.csv")
# 分析数据集信息
print("*" * 30 + " info " + "*" * 30)
print(titanic.info())
print(titanic.head())
输出：
****************************** info ******************************
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 891 entries, 0 to 890
Data columns (total 12 columns):
PassengerId 891 non-null int64
Survived 891 non-null int64
Pclass 891 non-null int64
Name 891 non-null object
Sex 891 non-null object
Age 714 non-null float64
SibSp 891 non-null int64
Parch 891 non-null int64
Ticket 891 non-null object
Fare 891 non-null float64
Cabin 204 non-null object
Embarked 889 non-null object
dtypes: float64(2), int64(5), object(5)
memory usage: 83.6+ KB
None
PassengerId Survived Pclass ... Fare Cabin Embarked
0 1 0 3 ... 7.2500 NaN S
1 2 1 1 ... 71.2833 C85 C
2 3 1 3 ... 7.9250 NaN S
3 4 1 1 ... 53.1000 C123 S
4 5 0 3 ... 8.0500 NaN S
[5 rows x 12 columns]
survived字段表示是否生存，我们以此作为预测目标。
y = titanic['survived']
print(y.head())
输出：
0 0
1 1
2 1
3 1
4 0
Name: survived, dtype: int64
我们取其中三个特征做分析演示，分别是：
- pclass：1-一等舱，2-二等舱，3-三等舱
- age年龄
- sex性别
x = titanic[['pclass', 'age', 'sex']]
print(x.info())
print(x.head())
输出：
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 891 entries, 0 to 890
Data columns (total 3 columns):
Pclass 891 non-null int64
Age 714 non-null float64
Sex 891 non-null object
dtypes: float64(1), int64(1), object(1)
memory usage: 21.0+ KB
None
Pclass Age Sex
0 3 22.0 male
1 1 38.0 female
2 3 26.0 female
3 1 35.0 female
4 3 35.0 male
缺失值处理
age字段存在缺失，用均值填充：
age_mean = x['age'].mean()
print("*" * 30 + " age_mean " + "*" * 30)
print(age_mean)
x['age'].fillna(age_mean, inplace=True)
print("*" * 30 + " 处理age缺失值后 " + "*" * 30)
print(x.info())
输出：
****************************** age_mean ******************************
29.69911764705882
****************************** 处理age缺失值后 ******************************
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 891 entries, 0 to 890
Data columns (total 3 columns):
Pclass 891 non-null int64
Age 891 non-null float64
Sex 891 non-null object
dtypes: float64(1), int64(1), object(1)
memory usage: 21.0+ KB
特征抽取 - onehot编码
为了方便使用字典特征抽取，构造字典列表：
x_dict_list = x.to_dict(orient='records')
print("*" * 30 + " train_dict " + "*" * 30)
print(pd.Series(x_dict_list[:5]))
dict_vec = DictVectorizer(sparse=False)
x = dict_vec.fit_transform(x_dict_list)
print("*" * 30 + " onehot编码 " + "*" * 30)
print(dict_vec.get_feature_names())
print(x[:5])
****************************** train_dict ******************************
0 {'Pclass': 3, 'Age': 22.0, 'Sex': 'male'}
1 {'Pclass': 1, 'Age': 38.0, 'Sex': 'female'}
2 {'Pclass': 3, 'Age': 26.0, 'Sex': 'female'}
3 {'Pclass': 1, 'Age': 35.0, 'Sex': 'female'}
4 {'Pclass': 3, 'Age': 35.0, 'Sex': 'male'}
dtype: object
****************************** onehot编码 ******************************
['Age', 'Pclass', 'Sex=female', 'Sex=male']
[[22. 3. 0. 1.]
[38. 1. 1. 0.]
[26. 3. 1. 0.]
[35. 1. 1. 0.]
[35. 3. 0. 1.]]
划分训练集和测试集
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.25)
决策树分类器
dec_tree = DecisionTreeClassifier()
dec_tree.fit(x_train, y_train)
print("*" * 30 + " 准确率 " + "*" * 30)
print(dec_tree.score(x_test, y_test))
输出：
****************************** 准确率 ******************************
0.7892376681614349
随机森林分类器
- n_jobs: -1表示设置为核心数量
- n_estimators: 决策树数目
- max_depth: 树最大深度
同时使用网格搜索最优超参数：
rf = RandomForestClassifier(n_jobs=-1)
param = {
"n_estimators": [120, 200, 300, 500, 800, 1200],
"max_depth": [5, 8, 15, 25, 30]
}
# 2折交叉验证
search = GridSearchCV(rf, param_grid=param, cv=2)
print("*" * 30 + " 超参数网格搜索 " + "*" * 30)
start_time = time.time()
search.fit(x_train, y_train)
print("耗时：{}".format(time.time() - start_time))
print("最优参数：{}".format(search.best_params_))
print("*" * 30 + " 准确率 " + "*" * 30)
print(search.score(x_test, y_test))
输出：
****************************** 超参数网格搜索 ******************************
耗时：66.85670185089111
最优参数：{'max_depth': 5, 'n_estimators': 120}
****************************** 准确率 ******************************
0.7847533632286996
最优的参数是{'max_depth': 5, 'n_estimators': 120}。
在我的2015款MacBookPro上，仅2折的交叉验证就跑了66秒。- -|
源码
Github: DecisionTree
