CODEWITHSUNDEEP
pythonpandas
user2805507
user2805507

Reputation:

extracting values from describe() in python

Is there any way we could extract values from describe function in python ?

for an example i have this table

basicprofiling = pdf.toPandas().describe()


        product   Bs_type     country   period  table_name
count       200     200       200       200     200
unique      2        1        1         1       2
top       Deposits  Retail  vietnam   daily    animal
freq        100     200     200        200      100

lets say i would want to extract and print product count,unique no and total type. is this achievable ?

this is what i tried basicprofiling.select('prf_product') but it is returning error on str not callable

Upvotes: 2

Views: 2996

Answers (3)

Matt L.
Matt L.

Reputation: 3631

Describe returns a DataFrame where the summary names are the index, so you can access all the counts (for example) using loc, like this:

import numpy as np
import pandas as pd

df = pd.DataFrame(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), columns=['a', 'b', 'c'])
data = df.describe()

data.loc['count']

And individual values like this:

data.loc["count","a"]

Upvotes: 1

Jeff Gruenbaum
Jeff Gruenbaum

Reputation: 403

If you run this line, you will see that print(type(df.describe())) actually returns a dataframe:

<class 'pandas.core.frame.DataFrame'>

So you can access values inside like you would a regular dataframe:

df = pd.DataFrame({"a":[1,4],"b":[2,1],"c":[7,9],"d":[1,3]})
print(df.describe()['a']['count'])

The output will be: 2.0

Upvotes: 0

Sheldore
Sheldore

Reputation: 39072

You can use to_frame() to get a DataFrame from the Series (output of describe) and then .T to tranaform the Series indices to column names. Then you can simply access the values you want. For example

s = pd.Series(['a', 'a', 'b', 'c'])
basicprofiling = s.describe().to_frame().T

print (basicprofiling['count'], basicprofiling['unique'])
# 4, 3

Upvotes: 0

Related Questions