0

I want to create multiple (two in this case) boxplots based on a data in a dataframe

I have the following dataframe:

 Country Fund R^2 Style 0 Austria BG EMCore Convertibles Global CHF R T 0.739131 Allocation 1 Austria BG EMCore Convertibles Global R T 0.740917 Allocation 2 Austria BG Trend A T 0.738376 Fixed Income 3 Austria Banken Euro Bond-Mix A 0.71161 Fixed Income 4 Austria Banken KMU-Fonds T 0.778276 Allocation 5 Brazil Banken Nachhaltigkeitsfonds T 0.912808 Allocation 6 Brazil Banken Portfolio-Mix A 0.857019 Allocation 7 Brazil Banken Portfolio-Mix T 0.868856 Fixed Income 8 Brazil Banken Sachwerte-Fonds T 0.730626 Fixed Income 9 Brazil Banken Strategie Wachstum T 0.918684 Fixed Income

I want to create a boxplot chart for each country summarized by Style and showing the distribution of R^2. I was thinking of groupby operation but somehow I don't manage to make two charts for each country.

Thanks in advance

Improve this question
2
  • how the data shell by grouped? Only by Country or by Country and Style? Commented Aug 13, 2019 at 10:36
  • I guess by country and style. For each country one boxplot chart consisting of two bars for style - because we have Allocation and Fixed Income. Hope this answers Commented Aug 13, 2019 at 10:40

3 Answers 3

Reset to default
2

Here You go. Description in code.

=^..^=

import pandas as pd import numpy as np import matplotlib.pyplot as plt from io import StringIO data = StringIO(""" Country R^2 Style Austria 0.739131 Allocation Austria 0.740917 Allocation Austria 0.738376 Fixed_Income Austria 0.71161 Fixed_Income Austria 0.778276 Allocation Brazil 0.912808 Allocation Brazil 0.857019 Allocation Brazil 0.868856 New_Style Brazil 0.730626 Fixed_Income Brazil 0.918684 Fixed_Income Brazil 0.618684 New_Style """) # load data into data frame df = pd.read_csv(data, sep=' ') # group data by Country grouped_data = df.groupby(['Country']) # create list of grouped data frames df_list = [] country_list = [] for item in list(grouped_data): df_list.append(item[1]) country_list.append(item[0]) # plot box for each Country for df in df_list: country = df['Country'].unique() df = df.drop(['Country'], axis=1) df = df[['Style', 'R^2']] columns_names = list(set(df['Style'])) # pivot rows into columns df = df.assign(g = df.groupby('Style').cumcount()).pivot('g','Style','R^2') # plot box df.boxplot(column=colums_names) plt.title(country[0]) plt.show()

Output:

enter image description here enter image description here

Improve this answer
Sign up to request clarification or add additional context in comments.

3 Comments

Thanks Zaraki. I think it will work. I didn't specify this but there are many countries and multiple styles - for example equity, fixed income, allocation etc. Therefore in df.boxplot(column=['Allocation', 'Fixed_Income', Equity', etc etc]) I input all the styles. Yet for some countries not all styles apply. So when the code finds a country with less styles than the specified in the list it gives an error. Do you know how I can tackle this. Maybe somehow to specify in df.boxplot(column=['Allocation', 'Fixed_Income' etc.]) that if some of the styles is not found to raise exception.
@MartinYordanovGeorgiev I updated my code with line: column_names. Now it should handle different styles.
Thanks Zaraki, works just fine. Much appreciated. Just corrected one typo in df.boxplot(column=colums_names) "n" is omitted from colums_names. I posted an alternative answer myself. You can check below if interested.
1

Came up with some solution myself. 

df= "This is the table from the original question" uniquenames=df.Country.unique() # create dictionary of the data with countries set as keys diction={elem:pd.DataFrame for elem in uniquenames} # fill dictionary with values for key in diction.keys(): diction[key]=df[:][df.Country==key] #plot the data for i in diction.keys(): diction[i].boxplot(column="R^2",by="Style", figsize=(15,6),patch_artist=True,fontsize=12) plt.xticks(rotation=90) plt.title(i,fontsize=12)
Improve this answer

Comments

0

Use seaborn for this kind of tasks. Here are a couple of options:

Use seaborn's boxplot

import seaborn as sns sns.set() # Note - the data is stored in a data frame df sns.boxplot(x='Country', y='R^2', hue='Style', data=df)

enter image description here

Alternatively, you can use seaborn's FacetGrid.

g = sns.FacetGrid(df, col="Country", row="Style") g = g.map(sns.boxplot, 'R^2', orient='v')

enter image description here

Improve this answer

Comments

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.