This notebook is an exercise in the Feature Engineering course. You can reference the tutorial at this link.

---

Introduction

介绍

In this exercise you'll start developing the features you identified in Exercise 2 as having the most potential. As you work through this exercise, you might take a moment to look at the data documentation again and consider whether the features we're creating make sense from a real-world perspective, and whether there are any useful combinations that stand out to you.

在本练习中，您将开始开发在练习 2 中确认的最有潜力的特征。 当您完成此练习时，您可能会花一些时间再次查看数据文档，并考虑我们正在创建的特征从现实世界的角度是否有意义，以及是否有任何对您来说有用的组合。

Run this cell to set everything up!

运行这个单元格来设置一切！

# Setup feedback system
from learntools.core import binder
binder.bind(globals())
from learntools.feature_engineering_new.ex3 import *

import numpy as np
import pandas as pd
from sklearn.model_selection import cross_val_score
from xgboost import XGBRegressor

def score_dataset(X, y, model=XGBRegressor()):
    # Label encoding for categoricals
    for colname in X.select_dtypes(["category", "object"]):
        X[colname], _ = X[colname].factorize()
    # Metric for Housing competition is RMSLE (Root Mean Squared Log Error)
    score = cross_val_score(
        model, X, y, cv=5, scoring="neg_mean_squared_log_error",
    )
    score = -1 * score.mean()
    score = np.sqrt(score)
    return score

# Prepare data
df = pd.read_csv("../input/fe-course-data/ames.csv")
X = df.copy()
y = X.pop("SalePrice")

---

Let's start with a few mathematical combinations. We'll focus on features describing areas -- having the same units (square-feet) makes it easy to combine them in sensible ways. Since we're using XGBoost (a tree-based model), we'll focus on ratios and sums.

让我们从一些数学组合开始。 我们将重点关注描述区域的特征 - 具有相同的单位（平方英尺）可以轻松地以合理的方式组合它们。 由于我们使用的是 XGBoost（基于树的模型），因此我们将重点关注比率和总和。

1) Create Mathematical Transforms

1) 创建数学变换

Create the following features:

创建以下特征：

• LivLotRatio: the ratio of GrLivArea to LotArea
• LivLotRatio：GrLivArea 与 LotArea 的比率
• Spaciousness: the sum of FirstFlrSF and SecondFlrSF divided by TotRmsAbvGrd
• Spaciousness：FirstFlrSF 和 SecondFlrSF 之和与 TotRmsAbvGrd的商
• TotalOutsideSF: the sum of WoodDeckSF, OpenPorchSF, EnclosedPorch, Threeseasonporch, and ScreenPorch
• TotalOutsideSF：WoodDeckSF、OpenPorchSF、EnclosurePorch、Threeseasonporch 和 ScreenPorch 的总和

# YOUR CODE HERE
X_1 = pd.DataFrame()  # dataframe to hold new features

#X_1["LivLotRatio"] = ____
#X_1["Spaciousness"] = ____
#X_1["TotalOutsideSF"] = ____

X_1["LivLotRatio"] = X.GrLivArea/X.LotArea
X_1["Spaciousness"] = (X.FirstFlrSF + X.SecondFlrSF)/X.TotRmsAbvGrd
X_1["TotalOutsideSF"] = X[["WoodDeckSF", "OpenPorchSF", "EnclosedPorch", "Threeseasonporch", "ScreenPorch"]].sum(axis=1)

# Check your answer
q_1.check()

Correct

# Lines below will give you a hint or solution code
#q_1.hint()
q_1.solution()

Solution:

X_1["LivLotRatio"] = df.GrLivArea / df.LotArea
X_1["Spaciousness"] = (df.FirstFlrSF + df.SecondFlrSF) / df.TotRmsAbvGrd
X_1["TotalOutsideSF"] = df.WoodDeckSF + df.OpenPorchSF + df.EnclosedPorch + df.Threeseasonporch + df.ScreenPorch

---

If you've discovered an interaction effect between a numeric feature and a categorical feature, you might want to model it explicitly using a one-hot encoding, like so:

如果您发现数字特征和分类特征之间存在交互效应，您可能希望使用 one-hot 编码对其进行显式建模，如下所示：

# One-hot encode Categorical feature, adding a column prefix "Cat"
# One-hot 编码分类特征，添加列前缀“Cat”
X_new = pd.get_dummies(df.Categorical, prefix="Cat")

# Multiply row-by-row
# 逐行相乘
X_new = X_new.mul(df.Continuous, axis=0)

# Join the new features to the feature set
# 将新特征加入到特征集中
X = X.join(X_new)

2) Interaction with a Categorical

2) 与分类的交互

We discovered an interaction between BldgType and GrLivArea in Exercise 2. Now create their interaction features.

我们在练习 2 中发现了BldgType和GrLivArea之间的交互。现在创建它们的交互特征。

# YOUR CODE HERE
# One-hot encode BldgType. Use `prefix="Bldg"` in `get_dummies`
#X_2 = ____ 
X_2 = pd.get_dummies(X["BldgType"], prefix="Bldg")
# Multiply
#X_2 = ____
X_2 = X_2.multiply(X["GrLivArea"], axis=0)

# Check your answer
q_2.check()

Correct

# Lines below will give you a hint or solution code
#q_2.hint()
q_2.solution()

Solution:

X_2 = pd.get_dummies(df.BldgType, prefix="Bldg")
X_2 = X_2.mul(df.GrLivArea, axis=0)

3) Count Feature

3) 计数特征

Let's try creating a feature that describes how many kinds of outdoor areas a dwelling has. Create a feature PorchTypes that counts how many of the following are greater than 0.0:

让我们尝试创建一个特征来描述住宅有多少种室外区域。 创建一个特征 PorchTypes 来计算以下有多少个大于 0.0：

WoodDeckSF
OpenPorchSF
EnclosedPorch
Threeseasonporch
ScreenPorch

X[["WoodDeckSF", "OpenPorchSF", "EnclosedPorch", "Threeseasonporch", "ScreenPorch"]].gt(0).sum(axis=1)

0       2
1       2
2       2
3       0
4       2
       ..
2925    1
2926    1
2927    2
2928    2
2929    2
Length: 2930, dtype: int64

X_3 = pd.DataFrame()

# YOUR CODE HERE
#X_3["PorchTypes"] = ____
X_3["PorchTypes"] = X[["WoodDeckSF", "OpenPorchSF", "EnclosedPorch", "Threeseasonporch", "ScreenPorch"]].gt(0).sum(axis=1)

# Check your answer
q_3.check()

Correct

# Lines below will give you a hint or solution code
#q_3.hint()
q_3.solution()

Solution:

X_3 = pd.DataFrame()

X_3["PorchTypes"] = df[[
    "WoodDeckSF",
    "OpenPorchSF",
    "EnclosedPorch",
    "Threeseasonporch",
    "ScreenPorch",
]].gt(0.0).sum(axis=1)

4) Break Down a Categorical Feature

4) 分解分类特征

MSSubClass describes the type of a dwelling:

MSSubClass 描述了住宅的类型：

df.MSSubClass.unique()

array(['One_Story_1946_and_Newer_All_Styles', 'Two_Story_1946_and_Newer',
       'One_Story_PUD_1946_and_Newer',
       'One_and_Half_Story_Finished_All_Ages', 'Split_Foyer',
       'Two_Story_PUD_1946_and_Newer', 'Split_or_Multilevel',
       'One_Story_1945_and_Older', 'Duplex_All_Styles_and_Ages',
       'Two_Family_conversion_All_Styles_and_Ages',
       'One_and_Half_Story_Unfinished_All_Ages',
       'Two_Story_1945_and_Older', 'Two_and_Half_Story_All_Ages',
       'One_Story_with_Finished_Attic_All_Ages',
       'PUD_Multilevel_Split_Level_Foyer',
       'One_and_Half_Story_PUD_All_Ages'], dtype=object)

You can see that there is a more general categorization described (roughly) by the first word of each category. Create a feature containing only these first words by splitting MSSubClass at the first underscore _. (Hint: In the split method use an argument n=1.)

您可以看到每个类别的第一个单词（粗略地）描述了更一般的分类。 通过在第一个下划线_处拆分MSSubClass，创建仅包含这些前几个单词的特征。 （提示：在split方法中使用参数n=1。）

X_4 = pd.DataFrame()

# YOUR CODE HERE
#____
X_4['MSClass'] = df["MSSubClass"].str.split("_", n=1, expand=True)[0]

# Check your answer
q_4.check()

Correct

# Lines below will give you a hint or solution code
#q_4.hint()
q_4.solution()

Solution:

X_4 = pd.DataFrame()

X_4["MSClass"] = df.MSSubClass.str.split("_", n=1, expand=True)[0]

5) Use a Grouped Transform

5) 使用分组变换

The value of a home often depends on how it compares to typical homes in its neighborhood. Create a feature MedNhbdArea that describes the median of GrLivArea grouped on Neighborhood.

房屋的价值通常取决于它与附近典型房屋的比较。 创建一个特征MedNhbdArea，描述按Neighborhood分组的GrLivArea的中位值。

X_5 = pd.DataFrame()

# YOUR CODE HERE
#X_5["MedNhbdArea"] = ____
X_5["MedNhbdArea"] = X.groupby("Neighborhood")["GrLivArea"].transform("median")
# Check your answer
q_5.check()

Correct

# Lines below will give you a hint or solution code
#q_5.hint()
q_5.solution()

Solution:

X_5 = pd.DataFrame()

X_5["MedNhbdArea"] = df.groupby("Neighborhood")["GrLivArea"].transform("median")

Now you've made your first new feature set! If you like, you can run the cell below to score the model with all of your new features added:

现在您已经制作了第一个新特征集！ 如果您愿意，您可以运行下面的单元格来对添加了所有新特征的模型进行评分：

X_new = X.join([X_1, X_2, X_3, X_4, X_5])
score_dataset(X_new, y)

0.13954039591355258

score_dataset(X, y)

0.14336777904947118

Keep Going

继续前进

Untangle spatial relationships by adding cluster labels to your dataset.

通过向数据集添加聚类标签来理清空间关系。