机器学习实战之一：一个完整的机器学习项目

一个机器学习项目主要步骤为：

1. 获取数据

• 使用Pandas加载数据，并返回一个包含所有数据的Pandas DataFrame对象。

   import pandas as pd

def load_housing_data(housing_path=HOUSING_PATH):
    csv_path = os.path.join(housing_path, "housing.csv")
    return pd.read_csv(csv_path)

• 使用DataFrame的 head()方法查看该数据集的前5行:
  
  housing.head()
• 使用 describe()方法展示数值属性的概括：
  
  housing.describe()
• 创建测试集（根据收入，进行分层采样）：

   from sklearn.model_selection import StratifiedShuffleSplit

split = StratifiedShuffleSplit(n_splits=1, test_size=0.2, random_state=42)

for train_index, test_index in split.split(housing, housing["income_cat"]):
    strat_train_set = housing.loc[train_index]
    strat_test_set = housing.loc[test_index]

2. 发现并可视化数据，发现规律

• 地理数据的可视化：

   housing.plot(kind="scatter", x="longitude", y="latitude", alpha=0.4,
    s=housing["population"]/100, label="population",
    c="median_house_value", cmap=plt.get_cmap("jet"), colorbar=True,
)
plt.legend()

数据的地理信息散点图

• 查找关联
  使用 corr()方法计算出每对属性间的标准相关系数（standard correlation coefficient，也称作皮尔逊相关系数）：

   >>> corr_matrix = housing.corr()
>>> corr_matrix["median_house_value"].sort_values(ascending=False)#每个属性和房价中位数的关联度
median_house_value    1.000000
median_income         0.687170
total_rooms           0.135231
housing_median_age    0.114220
households            0.064702
total_bedrooms        0.047865
population           -0.026699
longitude            -0.047279
latitude             -0.142826
Name: median_house_value, dtype: float64

• 尝试不同的属性组合

   >>> housing["rooms_per_household"] = housing["total_rooms"]/housing["households"]
>>> housing["bedrooms_per_room"] = housing["total_bedrooms"]/housing["total_rooms"]
>>> housing["population_per_household"]=housing["population"]/housing["households"]
>>> corr_matrix = housing.corr()
>>> corr_matrix["median_house_value"].sort_values(ascending=False)
median_house_value          1.000000
median_income               0.687170
rooms_per_household         0.199343
total_rooms                 0.135231
housing_median_age          0.114220
households                  0.064702
total_bedrooms              0.047865
population_per_household   -0.021984
population                 -0.026699
longitude                  -0.047279
latitude                   -0.142826
bedrooms_per_room          -0.260070
Name: median_house_value, dtype: float64
#可以看出来，与总房间数或卧室数相比，新的bedrooms_per_room属性与房价中位数的关联更强

3. 数据预处理

• 处理缺失值

   from sklearn.preprocessing import Imputer

imputer = Imputer(strategy="median")
housing_num = housing.drop("ocean_proximity", axis=1)#创建一份不包括文本属性ocean_proximity的数据副本
imputer.fit(housing_num)
X = imputer.transform(housing_num)

• 处理文本和类别属性(使用独热编码One-Hot Encoding)

   from sklearn.preprocessing import CategoricalEncoder # in future versions of Scikit-Learn

cat_encoder = CategoricalEncoder()
housing_cat_reshaped = housing_cat.values.reshape(-1, 1)
housing_cat_1hot = cat_encoder.fit_transform(housing_cat_reshaped)

• 特征缩放
  有两种常见的方法可以让所有的属性有相同的量度：线性函数归一化（Min-Max scaling）和标准化（standardization）。
• 转换流水线

   from sklearn.pipeline import FeatureUnion

num_attribs = list(housing_num)
cat_attribs = ["ocean_proximity"]

num_pipeline = Pipeline([
        ('selector', DataFrameSelector(num_attribs)),
        ('imputer', Imputer(strategy="median")),
        ('attribs_adder', CombinedAttributesAdder()),
        ('std_scaler', StandardScaler()),
    ])

cat_pipeline = Pipeline([
        ('selector', DataFrameSelector(cat_attribs)),
        ('cat_encoder', CategoricalEncoder(encoding="onehot-dense")),
    ])

full_pipeline = FeatureUnion(transformer_list=[
        ("num_pipeline", num_pipeline),
        ("cat_pipeline", cat_pipeline),
    ])

运行流水线 ：

   housing_prepared = full_pipeline.fit_transform(housing)

4. 选择模型，进行训练

• 线性回归模型

   from sklearn.linear_model import LinearRegression

lin_reg = LinearRegression()
lin_reg.fit(housing_prepared, housing_labels)

• 决策树模型

   from sklearn.tree import DecisionTreeRegressor

tree_reg = DecisionTreeRegressor()
tree_reg.fit(housing_prepared, housing_labels)

• 随机森林模型

   from sklearn.ensemble import RandomForestRegressor
forest_reg = RandomForestRegressor()
forest_reg.fit(housing_prepared, housing_labels)

使用 Scikit-Learn 的交叉验证功能---K 折交叉验证（K-fold cross-validation）：

   from sklearn.model_selection import cross_val_score
scores = cross_val_score(tree_reg, housing_prepared, housing_labels,
                         scoring="neg_mean_squared_error", cv=10)
rmse_scores = np.sqrt(-scores)

5. 微调模型

• 网格搜索

   from sklearn.model_selection import GridSearchCV

param_grid = [
    {'n_estimators': [3, 10, 30], 'max_features': [2, 4, 6, 8]},
    {'bootstrap': [False], 'n_estimators': [3, 10], 'max_features': [2, 3, 4]},
  ]

forest_reg = RandomForestRegressor()

grid_search = GridSearchCV(forest_reg, param_grid, cv=5,
                           scoring='neg_mean_squared_error')

grid_search.fit(housing_prepared, housing_labels)

• 随机搜索

6. 确定最优模型

完整项目代码: 一个完整的机器学习项目