Interactive boxplot with pandas and Jupyter notebook

Question

I have data in a pandas DataFrame, and I want to create an interactive boxplot that allows me to select the number of days, whilst plotting a boxplot for the values in each of the categories in the column 'category'

This is what my code/data looks like so far:

import numpy as np
import pandas as pd

categories=('A','B','C')

data = {
            'days':      np.random.randint(120, size=100), 
            'category':  np.random.choice(categories, 100),
            'value':     100.0 * np.random.random_sample(100)
       }

df = pd.DataFrame(data)

print(df)

   category  days      value
0         A     4  77.383981
1         A    31  63.011934
2         A     5   1.165061
3         C    59  23.588979
4         A    57  14.906734
5         C   106  33.366634
6         A    29  90.658570
7         B    25  16.137490
8         A   118  34.526302
9         C    76   4.111797
10        A    11  30.195917
..      ...   ...        ...
90        A    64  37.529774
91        A    76   3.771360
92        C   112  93.948775
93        C    14  34.855189
94        B    64  83.106007
95        A    10  78.346319
96        B    86  66.645889
97        A    46  12.969012
98        C    29  57.925427
99        A    59  34.526146

[100 rows x 3 columns]

I want to create a boxplot of the values for each of the categories (for a selected/specified number of days), with the different categories being plotted along the X-axis.

How do I do that using pandas (or matplotlib)?

(1) Can you go more into detail what "interactive plot in jupyter" would mean? As I see it, jupyter is already quite interactive, such that calling a cell with a new value creates a different plot. (2) What is number of days? A single value like `42`? How would like to select the number of days? — ImportanceOfBeingErnest, Jan 16 '17 at 00:06
@ImportanceOfBeingErnest: (1) I first came across this feature in Jupyter, by watching this video: https://www.youtube.com/watch?v=ZrRpN_IrcBA (concept was introduced around the 24:50 mark). (2) Number of days is an integer. Interactive widgets allow you to select the value from a dropdown box. — Homunculus Reticulli, Jan 16 '17 at 06:15
That would have been a piece of information to include in the question. See updated answer for how to implement this dropdown box. — ImportanceOfBeingErnest, Jan 16 '17 at 09:14

score 9 · Accepted Answer · edited Aug 29 '22 at 15:41

You can simply filter the dataframe by the number of days and then plot the respective boxplot.

numer_of_days = 42
df_filtered= df.loc[df['days'] < numer_of_days]  # use operators like ==, >=, <, etc.
df_filtered[["category", "value"]].boxplot( by="category", return_type='axes')

In order to get a dropdown field, you can use the `ipywidgets.interact()` function, to which you provide a function that plots the dataframe for that specific day. (In the following I restricted the number of days to 12, such that a dropdown actually makes sense for selecting a single day out of those.)

import numpy as np
import pandas as pd
from ipywidgets import interact
%matplotlib notebook

categories=('A','B','C')

data = {
            'days':      np.random.randint(12, size=100), 
            'category':  np.random.choice(categories, 100),
            'value':     100.0 * np.random.random_sample(100)
       }

df = pd.DataFrame(data)

def select_days(number_of_days):
    df_filtered= df.loc[df['days'] == int(number_of_days)] 
    ax = df_filtered[["category", "value"]].boxplot( by="category", return_type='axes')
    ax["value"].set_title("Day " + number_of_days)
    print(df_filtered)
    
days = [str(day) for day in np.arange(12)]

interact(select_days, number_of_days=days)

score 0 · Answer 2 · edited Oct 18 '20 at 18:34

0

How to display box, distribution and violin plot

f, axes = plt.subplots(5, 3, figsize=(20, 20))
colors = ["r", "g", "b", "m", "c"]

count = 0
for i in houseNumData:
    sb.boxplot(houseNumData[i], orient = "h", color = colors[count], ax = axes[count,0])
    sb.distplot(houseNumData[i], color = colors[count], ax = axes[count,1])
    sb.violinplot(houseNumData[i], color = colors[count], ax = axes[count,2])
    count += 1

edited Oct 18 '20 at 18:34

Joe Ferndz

8,417
2
13
33

answered Oct 18 '20 at 18:27

Terence Hwang

1
1

1

I recommend you explain your code so the OP can understand how to implement the solution into OP's problem statement. – Joe Ferndz Oct 18 '20 at 18:36

Interactive boxplot with pandas and Jupyter notebook

2 Answers2

Linked