数据挖掘项目实战—Kaggle入门竞赛：房价预测之EDA与特征工程

文章目录

一、房价预测比赛介绍
- 数据集介绍
二、EDA与特征工程

房价预测网址

一、房价预测比赛介绍

房价回归预测，依据一个房子的全方位信息，包括面积、地段、环境等79个变量来预测出房子的价格。你的工作是预测每栋房子的销售价格。对于测试集中的每个Id，必须预测SalePrice变量的值。
这个比赛要求使用的metric是Root-Mean-Squared-Error (RMSE) between the logarithm of the predicted value and the logarithm of the observed sales price即观测到销售价格的对数与预测价格的对数之间的RMSE
提交文件的格式如下：

Id,SalePrice
1461,169000.1
1462,187724.1233
1463,175221
etc.

数据集介绍

一共有4个文件：train.csv是训练文件，test.csv是测试文件，data_description.txt特征描述文件，sample_submission.csv提交文件样例
train.csv文件字段含义：(英语水平有限，机翻的很多没那味)
SalePrice - 目标变量，销售价格
MSSubClass: 特征变量，建筑等级，应该属于类别型变量，但是数据中是以连续型数据表示的
MSZoning: 特征变量，总分区分类，类别型变量
LotFrontage: 特征变量，房子周围的街道数（有缺失值）
LotArea: 特征变量，土地面积
Street: 特征变量，道路类型（都是Pave）
Alley: 特征变量，巷道通道类型
LotShape: 特征变量，财产类型General shape of property
LandContour:特征变量，财产状况 Flatness of the property
Utilities: 特征变量，公共设施，都是(AllPub)Type of utilities available
LotConfig: 特征变量，配置位置Lot configuration
LandSlope: 特征变量，土地坡度，类别型变量
Neighborhood: 特征变量，在艾姆斯城市范围内的物理位置Physical locations within Ames city limits
Condition1: 特征变量，靠近主要公路或铁路Proximity to main road or railroad
Condition2: 特征变量，靠近主要公路或铁路（如果有第二条的话）Proximity to main road or railroad (if a second is present)
BldgType: 特征变量，住宅类型Type of dwelling
HouseStyle: 特征变量，住宅的风格Style of dwelling
OverallQual: 特征变量，整体的材料与质量Overall material and finish quality
OverallCond: 特征变量，总体状况评级Overall condition rating
YearBuilt: 特征变量，原始的施工日期Original construction date
YearRemodAdd: 特征变量，改变日期Remodel date
RoofStyle: 特征变量，屋顶的类型Type of roof
RoofMatl: 特征变量，屋顶的材料Roof material
Exterior1st: 特征变量，房屋的外墙Exterior covering on house
Exterior2nd: 特征变量，房屋外部的覆盖物Exterior covering on house (if more than one material)
MasVnrType: 特征变量，表层砌体类型Masonry veneer type
MasVnrArea: 特征变量，砌筑单板面积(平方英尺)Masonry veneer area in square feet
ExterQual: 特征变量，外部材质Exterior material quality
ExterCond: 特征变量，材料表面的现状Present condition of the material on the exterior
Foundation: 特征变量，基础类型Type of foundation
BsmtQual: 特征变量，地下室的高度Height of the basement
BsmtCond: 特征变量，地下室的一般状况General condition of the basement
BsmtExposure: 特征变量，室外或花园水平的地下室墙壁Walkout or garden level basement walls
BsmtFinType1: 特征变量，地下室完工区域质量Quality of basement finished area
BsmtFinSF1: 特征变量，类型1完成平方英尺Type 1 finished square feet
BsmtFinType2: 特征变量，第二成品区质量(如有)Quality of second finished area (if present)
BsmtFinSF2: 特征变量，类型2完成平方英尺Type 2 finished square feet
BsmtUnfSF: 特征变量，未完工的地下室面积Unfinished square feet of basement area
TotalBsmtSF: 特征变量，地下室面积总计平方英尺Total square feet of basement area
Heating: 特征变量，加热类型Type of heating
HeatingQC: 特征质量，加热质量和条件Heating quality and condition
CentralAir: 特征变量，中央空调Central air conditioning
Electrical: 特征变量，电气系统Electrical system
1stFlrSF: 特征变量，一楼的平方英尺First Floor square feet
2ndFlrSF: 特征变量，二楼的平方英尺Second floor square feet
LowQualFinSF: 特征变量，低质量的成品平方英尺(所有楼层)Low quality finished square feet (all floors)
GrLivArea: 特征变量，地面以上居住面积平方英尺Above grade (ground) living area square feet
BsmtFullBath: 特征变量，地下室全套浴室Basement full bathrooms
BsmtHalfBath: 特征变量，半地下室卫生间Basement half bathrooms
FullBath: 特征变量，地上全浴室Full bathrooms above grade
HalfBath: 特征变量，地上半浴Half baths above grade
Bedroom: 特征变量，非地下室卧室数量Number of bedrooms above basement level
Kitchen: 特征变量，厨房的数量Number of kitchens
KitchenQual: 特征变量，厨房的质量Kitchen quality
TotRmsAbvGrd: 特征变量，房间总数(不含浴室)Total rooms above grade (does not include bathrooms)
Functional: 特征变量，家庭功能评级Home functionality rating
Fireplaces: 特征变量，壁炉的数量Number of fireplaces
FireplaceQu: 特征变量，壁炉的质量Fireplace quality
GarageType: 特征变量，出库的位置Garage location
GarageYrBlt: 特征变量，车库建成的年份Year garage was built
GarageFinish: 特征变量，车库的内部装修Interior finish of the garage
GarageCars: 特征变量，按汽车容量计算车库的大小Size of garage in car capacity
GarageArea: 特征变量，车库大小(以平方英尺为单位)Size of garage in square feet
GarageQual: 特征变量，车库的质量Garage quality
GarageCond: 特征变量，车库的条件Garage condition
PavedDrive: 特征变量，道路车道Paved driveway
WoodDeckSF: 特征变量，以平方英尺为单位计算木甲板面积Wood deck area in square feet
OpenPorchSF: 特征变量，开放式门廊面积Open porch area in square feet
EnclosedPorch: 特征变量，以平方英尺为单位的封闭门廊区域Enclosed porch area in square feet
3SsnPorch: 特征变量，三季门廊面积平方英尺Three season porch area in square feet
ScreenPorch: 特征变量，屏风门廊面积(平方英尺)Screen porch area in square feet
PoolArea: 特征变量，泳池面积(平方英尺)Pool area in square feet
PoolQC: 特征变量，泳池的质量Pool quality
Fence: 特征变量，栅栏质量Fence quality
MiscFeature: 特征变量，其他类别中未涉及的杂项特性Miscellaneous feature not covered in other categories
MiscVal: 特征变量，杂项功能的价值$Value of miscellaneous feature
MoSold: 特征变量，月销售Month Sold
YrSold: 特征变量，年销售Year Sold
SaleType: 特征变量，销售类型Type of sale
SaleCondition: 特征变量，销售的条件Condition of sale

二、EDA与特征工程

没有在本地跑，直接在kaggle上做的，代码见网址：

kaggle房价预测—数据预处理与特征工程

如果有什么好的建议，欢迎评论区留言！