datatable.nunique()

datatable.

nunique

(

cols

)

sourcetests

Added in version 1.1.0

Count the number of unique values for each column from cols.

Parameters

cols

FExpr

Input columns.

return

Expr

f-expression having one row, and the same names and number of columns as in cols. All the returned column stypes are int64.

except

TypeError

The exception is raised when one of the columns from cols has an unsupported type, e.g. obj64.

Examples

from datatable import dt, f df = dt.Frame({'A': [1, 1, 2, None, 1, 2], 'B': [None, 2, 3, 4, None, 5], 'C': [1, 2, 1, 1, 2, 4]}) df

	A	B	C
	int32	int32	int32
0	1	NA	1
1	1	2	2
2	2	3	1
3	NA	4	1
4	1	NA	2
5	2	5	4

6 rows × 3 columns

Count the number of unique values for each column in the frame:

df[:, dt.nunique(f[:])]

	A	B	C
	int64	int64	int64
0	2	4	3

1 row × 3 columns

Count the number of unique values for column B only:

df[:, dt.nunique(f.B)]

	B
	int64
0	4

1 row × 1 column