Level 1 Pandas 검색/색인

2019-10-07
Table: order
id user_id product_id date price address state
1 3 9 2017-01-01 500 Seoul confirmed
2 1 7 2017-01-03 700 Seoul confirmed
3 3 8 2017-01-03 900 Daejeon confirmed
4 4 2 2017-01-07 500   canceled
5 7 3 2017-01-09 700 Incheon confirmed
6 5 7 2017-01-09 600 Busan canceled
7 2 5 2017-01-10 200   canceled
Column 검색
order["date"]
order[["user_id", "date", "amount"]]

columns = ["user_id", "date", "amount"]
order[columns]
Row 검색
# loc = locate
order.loc
order.loc[1]
order.loc[[1, 3, 7]]

order_ids = [1, 3, 7]
order.loc[order_ids]

type(order.loc[[1, 3, 7]])
# pandas.core. frame.DataFrame
Row/Column 검색
order_date = order.loc[1]["date"]
order_date

%timeit order.loc[1]["date"]

order_date = order.loc[1, "date"]
order_date

%timeit order.loc[1, "date"]
Row/Column 검색: at, loc
  • pandas at이 loc보다 속도가 빠름
  • at은 row, column 모두 하나씩만 접근 가능
order.at[1, "date"]
색인
order["date"] == "2017-01-03"
id  
1 False
2 True
3 True
4 False 
order[order["date"] == "2017-01-03"]
  user_id product_id date price address state
id            
2 1 7 2017-01-03 700 Seoul confirmed
3 3 8 2017-01-03 900 Daejeon confirmed 
order["price"] >= 500
id  
1 True
2 True
3 True 
order[order["price"] >= 500]
  user_id product_id date price address state
id            
1 3 9 2017-01-01 500 Seoul confirmed
2 1 7 2017-01-03 700 Seoul confirmed
다중조건
(order["price"] >= 500) & (order["state"] == "confirmed")
id  
1 True
2 True 
order[(order["price"] >= 500) & (order["state"] == "confirmed")]
id  
1 True
2 True 
high = (order["price"] >= 500)
confirmed = (order["state"] == "confirmed")
order[high & confirmed]
id  
1 True
2 True 
order.loc[order["date"] == "2017-01-09", "price"]

order.loc[order["date"] == "2017-01-09", ["date", "price"]]