The report presents the profile of each column as one row of a table: the column’s name and type, the number and fraction of missing and distinct values, and its statistics. It’s the quickest way to get an overview of a table you haven’t seen before, and it’s what col_summary_tbl() returns.
Each row starts with an icon for the column’s type. Statistics that don’t apply to any column are left out, and those that don’t apply to a particular column are shown as -. Statistics for string columns describe the lengths of the strings (as the note below the table says). Boolean columns show the percentages of True and False values in place of the number of distinct values. Columns of unsupported types are listed with an unsupported type badge and no statistics.
Parameters
show_sample_data:bool=False
Should a column with the first five non-missing values of each column be added? This is off by default, since the values can be long, and the table may hold sensitive data.
Returns
GT
A Great Tables object.
Examples
Let’s scan the game_revenue dataset and look at the report.
import pointblank as pbgame_revenue = pb.load_dataset(dataset="game_revenue", tbl_type="polars")scan = pb.DataScan(game_revenue, tbl_name="game_revenue")scan.get_tabular_report()
Polarsgame_revenueRows2,000Columns11
Column
NA
UQ
Mean
SD
Min
P5
Q1
Med
Q3
P95
Max
IQR
player_id
String
0 0
76 0.04
15
0
15
15
15
15
15
15
15
0
session_id
String
0 0
618 0.31
24
0
24
24
24
24
24
24
24
0
session_start
Datetime(time_unit='us', time_zone='UTC')
0 0
618 0.31
-
-
2015 01 01 01:31:03+00:00
-
-
-
-
-
2015 01 21 03:59:23+00:00
-
time
Datetime(time_unit='us', time_zone='UTC')
0 0
1989 0.99
-
-
2015 01 01 01:31:27+00:00
-
-
-
-
-
2015 01 21 04:06:29+00:00
-
item_type
String
0 0
2 <.01
2.22
0.42
2
2
2
2
2
3
3
0
item_name
String
0 0
24 0.01
7.35
1.27
5
5
7
8
8
9
11
1
item_revenue
Float64
0 0
508 0.25
4.34
13.21
0.004
0.01
0.09
0.38
1.25
21.99
142.99
1.16
session_duration
Float64
0 0
310 0.15
25.63
9.55
3.2
4.2
18.5
26.5
33.82
39.5
41.0
15.32
start_day
Date
0 0
20 0.01
-
-
2015 01 01
-
-
-
-
-
2015 01 21
-
acquisition
String
0 0
6 <.01
7.97
2.54
5
5
7
7
8
14
14
1
country
String
0 0
23 0.01
8.53
3.13
5
5
6
7
12
13
14
6
String columns statistics regard the string's length.
The header shows the table type and its dimensions. In the rows, item_revenue has a long right tail (its 95th percentile is far below its maximum), and the string columns player_id and session_id have fixed lengths. To see what the values look like, add the sample data:
String columns statistics regard the string's length.
The samples make it clear that player_id and session_id are codes, and they show the format of each column’s values. The report is a Great Tables object, so it can be saved to a standalone HTML file with its write_raw_html() method.