Question

我正在尝试将两个csv文件（客户购买数据，产品数据）作为数据框读取，然后进行组合和转动。

示例：

Customer Purchase Data:
CustomerID ProductId
1          39
1          6
2          8
3          39
3          40

Product Data:
ProductId Name
6         Car
8         House
39        Plane
40        Boat

Desired Pivot Table
ProductId Name  Cust_1 Cust_2 Cust_3
6         Car   1      0      0
8         House 0      1      0
39        Plane 1      0      1
40        Boat  0      0      1

我的问题是：可以这样做吗？
应该这样做吗？我可以在Excel中将其转换为csv。

Answer 1

这是另外两个步骤。

第1步：加入两个表

using DataFrames

### Create the DataFrame
customer = DataFrame(customerid = [1, 1, 2, 3, 3],
                     productid = [39, 6, 8, 39, 40])

product = DataFrame(productid = [6, 8, 39, 40],
                    name = ["Car", "House", "Plane", "Boat"])


res = join(customer, product, on = :productid)
# 5x3 DataFrames.DataFrame
# | Row | customerid | productid | name    |
# |-----|------------|-----------|---------|
# | 1   | 1          | 6         | "Car"   |
# | 2   | 2          | 8         | "House" |
# | 3   | 1          | 39        | "Plane" |
# | 4   | 3          | 39        | "Plane" |
# | 5   | 3          | 40        | "Boat"  |

第2步：：使用＆＃34; 1＆＃34;添加虚拟列并取消堆叠DataFrame（从长格式移动到宽格式）

### Add dummy column
res[:tmp] = 1
res
# 5x4 DataFrames.DataFrame
# | Row | customerid | productid | name    | tmp |
# |-----|------------|-----------|---------|-----|
# | 1   | 1          | 6         | "Car"   | 1   |
# | 2   | 2          | 8         | "House" | 1   |
# | 3   | 1          | 39        | "Plane" | 1   |
# | 4   | 3          | 39        | "Plane" | 1   |
# | 5   | 3          | 40        | "Boat"  | 1   |


### Pivot from long to Wide
res = unstack(res, :customerid, :tmp)
# 4x5 DataFrames.DataFrame
# | Row | productid | name    | 1  | 2  | 3  |
# |-----|-----------|---------|----|----|----|
# | 1   | 6         | "Car"   | 1  | NA | NA |
# | 2   | 8         | "House" | NA | 1  | NA |
# | 3   | 39        | "Plane" | 1  | NA | 1  |
# | 4   | 40        | "Boat"  | NA | NA | 1  |


### Finally we can replace NA by 0
[res[isna(res[col]), col] = 0 for col in [symbol("1"), 
                                          symbol("2"), 
                                          symbol("3")]]
res
# 4x5 DataFrames.DataFrame
# | Row | productid | name    | 1 | 2 | 3 |
# |-----|-----------|---------|---|---|---|
# | 1   | 6         | "Car"   | 1 | 0 | 0 |
# | 2   | 8         | "House" | 0 | 1 | 0 |
# | 3   | 39        | "Plane" | 1 | 0 | 1 |
# | 4   | 40        | "Boat"  | 0 | 0 | 1 |

如果要更改列名，可以手动执行

names!(res, [:productid, :name, :cust_1, :cust_2, :cust_3])

Answer 2

你可以。您可以DataFrames.jl使用join：

using DataFrames
cp = readtable("data/Customer_Purchase_Data.csv", separator = ' ')
p  = readtable("data/Product_Data.csv", separator = ' ')

f = join(cp, p, on = :ProductId)

5x3 DataFrames.DataFrame
| Row | CustomerID | ProductId | Name    |
|-----|------------|-----------|---------|
| 1   | 1          | 6         | "Car"   |
| 2   | 2          | 8         | "House" |
| 3   | 1          | 39        | "Plane" |
| 4   | 3          | 39        | "Plane" |
| 5   | 3          | 40        | "Boat"  |

与Julia合并和透视数据框架

2 个答案: