How to deselect pydatatable columns based on their types?

Question

I have created a datatable as,

DT_X = dt.Frame({'x':[1,2,3,4,5],
          'y':[0.1,0.5,0.9,1.5,4.3],
          'z':['a','b','c','d','e'],
          'u':[True,False,True,False,False],
          'v':[10,20,30,40,50],
          'r':[4.5,5.5,6.5,8.5,10.5]
         })

and its output as

Out[4]: 
   |  x    y  z    u   v     r
-- + --  ---  --  --  --  ----
 0 |  1  0.1  a    1  10   4.5
 1 |  2  0.5  b    0  20   5.5
 2 |  3  0.9  c    1  30   6.5
 3 |  4  1.5  d    0  40   8.5
 4 |  5  4.3  e    0  50  10.5

[5 rows x 6 columns]

The datatable types can be checked as,

In [5]: DT_X.stypes                                                                                                                                                                                         
Out[5]: 
(stype.int32,
 stype.float64,
 stype.str32,
 stype.bool8,
 stype.int32,
 stype.float64)

I can now select integer and bool type columns as

DT_X[:,[dt.int32,dt.bool8]]

Out[6]: 
   |  x   v   u
-- + --  --  --
 0 |  1  10   1
 1 |  2  20   0
 2 |  3  30   1
 3 |  4  40   0
 4 |  5  50   0

[5 rows x 3 columns]

As recommended the columns can be deselected with this code,

DT_X[:,f[:].remove([f.x,f.v,f.u])]

Out[7]: 
   |   y  z      r
-- + ---  --  ----
 0 | 0.1  a    4.5
 1 | 0.5  b    5.5
 2 | 0.9  c    6.5
 3 | 1.5  d    8.5
 4 | 4.3  e   10.5

[5 rows x 3 columns]

So, How to deselect only integer and bool type columns ?. the below code chunks is not worked out

DT_X[:,f[:].remove([dt.int32,dt.bool8])]

score 3 · Accepted Answer · answered Jun 19 '20 at 14:37

use the f-expression to wrap the types you want to remove :

DT_X[:,f[:].remove([f[int],f[bool]])]

    y   z   r
▪▪▪▪▪▪▪▪    ▪▪▪▪    ▪▪▪▪▪▪▪▪
0   0.1 a   4.5
1   0.5 b   5.5
2   0.9 c   6.5
3   1.5 d   8.5
4   4.3 e   10.5

For your code above, you need to wrap the data types with the f-expression :

 DT_X[:,f[:].remove([f[dt.int32], f[dt.bool8]])]

score 1 · Answer 2 · answered Jun 19 '20 at 13:40

I have figured out a workaround for this.

First i collected the required column names and created a f expression on it.

fil_cols = [ f[col] for col in list(DT_X[:,[dt.int32,dt.bool8]].names)]

and apply this filter on datatable as,

DT_X[:,f[:].remove(fil_cols)]

and it generates an expected output.

Out[5]: 
   |   y  z      r
-- + ---  --  ----
 0 | 0.1  a    4.5
 1 | 0.5  b    5.5
 2 | 0.9  c    6.5
 3 | 1.5  d    8.5
 4 | 4.3  e   10.5

[5 rows x 3 columns]

How to deselect pydatatable columns based on their types?

2 Answers2