擷取特徵

使用 PySpark 進行特徵工程

John Hogue

Lead Data Scientist, General Mills

用文字比對擷取年限

ROOF
Asphalt Shingles, Pitched, Age 8 Years or Less
Asphalt Shingles, Age Over 8 Years
Asphalt Shingles, Age 8 Years or Less
Asphalt Shingles

Roof_Age	轉為	Roof>8yrs
Age 8 Years or Less	⇨	0
Age Over 8 Years	⇨	1
Age 8 Years or Less	⇨	0
NULL	⇨	NULL

用文字比對擷取年限

from pyspark.sql.functions import when

# 建立布林篩選條件
find_under_8 = df['ROOF'].like('%Age 8 Years or Less%')
find_over_8 = df['ROOF'].like('%Age Over 8 Years%')

# 用 when() 和 otherwise() 套用篩選
df = df.withColumn('old_roof', (when(find_over_8, 1)
                               .when(find_under_8, 0)
                               .otherwise(None)))

# 檢視結果
df[['ROOF', 'old_roof']].show(3, truncate=100)

+----------------------------------------------+--------+
|                                          ROOF|old_roof|
+----------------------------------------------+--------+
|                                          null|    null|
|Asphalt Shingles, Pitched, Age 8 Years or Less|       0|
|            Asphalt Shingles, Age Over 8 Years|       1|
+----------------------------------------------+--------+
only showing top 3 rows

分割欄位

ROOF	轉為	Roof_Material
Asphalt Shingles, Pitched, Age 8 Years or Less	⇨	Asphalt Shingles
Null	⇨
Asphalt Shingles, Age Over 8 Years	⇨	Asphalt Shingles
Metal, Age 8 Years or Less	⇨	Metal
Tile, Age 8 Years or Less	⇨	Tile
Asphalt Shingles	⇨	Asphalt Shingles

分割欄位

from pyspark.sql.functions import split

# 依逗號分割欄位為清單
split_col = split(df['ROOF'], ',')

# 取清單第 1 個值放入新欄位
df = df.withColumn('Roof_Material', split_col.getItem(0))

# 檢視結果
df[['ROOF', 'Roof_Material']].show(5, truncate=100)

+----------------------------------------------+----------------+
|                                          ROOF|   Roof_Material|
+----------------------------------------------+----------------+
|                                          null|            null|
|Asphalt Shingles, Pitched, Age 8 Years or Less|Asphalt Shingles|
|                                          null|            null|
|Asphalt Shingles, Pitched, Age 8 Years or Less|Asphalt Shingles|
|            Asphalt Shingles, Age Over 8 Years|Asphalt Shingles|
+----------------------------------------------+----------------+
only showing top 5 rows

Explode！

起始紀錄

NO	roof_list
2	[Asphalt Shingles, Pitched, Age 8 Years or Less]

展開後的紀錄

NO	ex_roof_list
2	Asphalt Shingles
2	Pitched
2	Age 8 Years or Less

Pivot！

展開後的紀錄

NO	ex_roof_list
2	Asphalt Shingles
2	Pitched
2	Age 8 Years or Less

樞紐分析後的紀錄

NO	Age 8 Years or Less	Age Over 8 Years	Asphalt Shingles	Flat	Metal	Other	Pitched	...
2	0	1	1	0	0	0	1	...

Explode 與 Pivot！

from pyspark.sql.functions import split, explode, lit, coalesce, first

# 依逗號分割欄位為清單
df = df.withColumn('roof_list', split(df['ROOF'], ', '))

# 將清單展開為多筆紀錄（每個值一筆）
ex_df = df.withColumn('ex_roof_list', explode(df['roof_list']))

# 建立常數虛擬欄位
ex_df = ex_df.withColumn('constant_val', lit(1))

# 將值樞紐化為布林欄位
piv_df = ex_df.groupBy('NO').pivot('ex_roof_list')\
  .agg(coalesce(first('constant_val')))

來整理一些特徵吧！

使用 PySpark 進行特徵工程