Using dtale¶

dtale is "visualizer for pandas data structures" that can be used in Jupyter notebooks. It's fancy!

Installing dtale¶

Installing dtale just requires a pip install:

pip install dtale

It takes a while to install, but that's it!

How to display and sort with dtale¶

Using qgrid to display and sort your data makes life easy. All you need to do is run qgrid.show_grid and feed it your dataframe.

In this case we're using precision=2 to round life expectancy to two decimal places.

In [ ]:

import pandas as pd
import dtale

df = pd.read_csv("countries.csv")
dtale.show(df)

import pandas as pd import dtale df = pd.read_csv("countries.csv") dtale.show(df)

You can't see the interactive version on the web, so here's a screenshot.

How to edit with dtale¶

dtale is also a great took for making manual edits! The edits are saved and can be used later.

First, we'll start with a pandas dataframe. This one is a few science fiction authors.

In [5]:

import pandas as pd

df = pd.DataFrame([
    { 'name': 'Philip K. Dick', 'book': 'Do Androids Dream of Electric Sheep?' },
    { 'name': 'Ursula K. Le Guin', 'book': 'The Left Hand of Darkness' },
    { 'name': 'Stanisław Lem', 'book': 'Solaris' }
])

df.head()

import pandas as pd df = pd.DataFrame([ { 'name': 'Philip K. Dick', 'book': 'Do Androids Dream of Electric Sheep?' }, { 'name': 'Ursula K. Le Guin', 'book': 'The Left Hand of Darkness' }, { 'name': 'Stanisław Lem', 'book': 'Solaris' } ]) df.head()

Out[5]:

	name	book
0	Philip K. Dick	Do Androids Dream of Electric Sheep?
1	Ursula K. Le Guin	The Left Hand of Darkness
2	Stanisław Lem	Solaris

While Stanisław Lem is great, not everyone knows how to type an ł – we'll use dtale to change his first name from Stanisław to Stanislaw.

Unlike when we're just sorting and filtering, this time we're saving our spreadsheet as a variable named d.

In [ ]:

import dtale

d = dtale.show(df)
d

import dtale d = dtale.show(df) d

(Pretend that you're watching me edit this the ł to an l in Stanisław's name)

Edits made in dtale do not edit your original dataframe. In order to get the edited dataframe, you ask dtale for the changed df with d.data.

We use .copy() to make sure that future edits don't affect our dataframe.

In [12]:

edited_df = d.data.copy()
edited_df

edited_df = d.data.copy() edited_df

Out[12]:

	name	book
0	Philip K. Dick	Do Androids Dream of Electric Sheep?
1	Ursula K. Le Guin	The Left Hand of Darkness
2	Stanislaw Lem	Solaris

Comparing df before and after edits¶

It's probably important to know what was edited. While unfortunately dtale doesn't give us a nice JSON diff or anything, we have a couple simple options to see what edits were made.

Compare each row with df.merge¶

In this situation, we use indicator=True to mark whether the row is found in the original or edited version of the data. You're probably interested in left_only and right_only.

In [13]:

df.merge(edited_df, indicator=True, how='outer')

df.merge(edited_df, indicator=True, how='outer')

Out[13]:

	name	book	_merge
0	Philip K. Dick	Do Androids Dream of Electric Sheep?	both
1	Ursula K. Le Guin	The Left Hand of Darkness	both
2	Stanisław Lem	Solaris	left_only
3	Stanislaw Lem	Solaris	right_only

Only show the changes¶

You can also use df.compare to just illustrate the changes between the two.

In [14]:

edits = df.compare(edited_df)
edits

edits = df.compare(edited_df) edits

Out[14]:

	name
	self	other
2	Stanisław Lem	Stanislaw Lem

Shutting down dtale¶

dtale is a magic little service running in the background. To shut it down you use d.kill()