[FIXED] Plot multiple distplot in seaborn Facetgrid

Issue

I have a dataframe which looks like below:

df:

RY         MAJ_CAT                  Value
2016    Cause Unknown              0.00227
2016    Vegetation                 0.04217
2016    Vegetation                 0.04393
2016    Vegetation                 0.07878
2016    Defective Equip            0.00137
2018    Cause Unknown              0.00484
2018    Defective Equip            0.01546
2020    Defective Equip            0.05169
2020    Defective Equip            0.00515
2020    Cause Unknown              0.00050

I want to plot the distribution of the value over the given years. So I used distplot of seaborn by using following code:

year_2016 = df[df['RY']==2016]
year_2018 = df[df['RY']==2018]
year_2020 = df[df['RY']==2020]
sns.distplot(year_2016['value'].values, hist=False,rug=True)    
sns.distplot(year_2018['value'].values, hist=False,rug=True)   
sns.distplot(year_2020['value'].values, hist=False,rug=True)

In the next step I want to plot the same value distribution over the given year w.r.t MAJ_CAT. So I decided to use Facetgrid of seaborn, below is the code :

g = sns.FacetGrid(df,col='MAJ_CAT')
g = g.map(sns.distplot,df[df['RY']==2016]['value'].values, hist=False,rug=True))    
g = g.map(sns.distplot,df[df['RY']==2018]['value'].values, hist=False,rug=True))    
g = g.map(sns.distplot,df[df['RY']==2020]['value'].values, hist=False,rug=True))

However, when it ran the above command, it throws the following error:

 KeyError: "None of [Index([(0.00227, 0.04217, 0.043930000000000004, 0.07877999999999999, 0.00137, 0.0018800000000000002, 0.00202, 0.00627, 0.00101, 0.07167000000000001, 0.01965, 0.02775, 0.00298, 0.00337, 0.00088, 0.04049, 0.01957, 0.01012, 0.12065, 0.23699, 0.03639, 0.00137, 0.03244, 0.00441, 0.06748, 0.00035, 0.0066099999999999996, 0.00302, 0.015619999999999998, 0.01571, 0.0018399999999999998, 0.03425, 0.08046, 0.01695, 0.02416, 0.08975, 0.0018800000000000002, 0.14743, 0.06366000000000001, 0.04378, 0.043, 0.02997, 0.0001, 0.22799, 0.00611, 0.13960999999999998, 0.38871, 0.018430000000000002, 0.053239999999999996, 0.06702999999999999, 0.14103, 0.022719999999999997, 0.011890000000000001, 0.00186, 0.00049, 0.13947, 0.0067, 0.00503, 0.00242, 0.00137, 0.00266, 0.38638, 0.24068, 0.0165, 0.54847, 1.02545, 0.01889, 0.32750999999999997, 0.22526, 0.24516, 0.12791, 0.00063, 0.0005200000000000001, 0.00921, 0.07665, 0.00116, 0.01042, 0.27046, 0.03501, 0.03159, 0.46748999999999996, 0.022090000000000002, 2.2972799999999998, 0.69021, 0.22529000000000002, 0.00147, 0.1102, 0.03234, 0.05799, 0.11744, 0.00896, 0.09556, 0.03202, 0.01347, 0.00923, 0.0034200000000000003, 0.041530000000000004, 0.04848, 0.00062, 0.0031100000000000004, ...)], dtype='object')] are in the [columns]"

I am not sure where am I making the mistake. Could anyone please help me in fixing the issue?

Solution

setup the dataframe

import pandas as pd
import numpy as np
import seaborn as sns

# setup dataframe of synthetic data
np.random.seed(365)
data = {'RY': np.random.choice([2016, 2018, 2020], size=400),
        'MAJ_CAT': np.random.choice(['Cause Unknown', 'Vegetation', 'Defective Equip'], size=400),
        'Value': np.random.random(size=400) }

df = pd.DataFrame(data)

Updated Answer

• From seaborn v0.11
• Use sns.displot with kind='kde' and rug=True
  • Is a figure-level interface for drawing distribution plots onto a FacetGrid.

Plotting all 'MAJ_CAT' together

sns.displot(data=df, x='Value', hue='RY', kind='kde', palette='tab10', rug=True)

Plotting 'MAJ_CAT' separately

sns.displot(data=df, col='MAJ_CAT', x='Value', hue='RY', kind='kde', palette='tab10', rug=True)

Original Answer

• In seaborn v0.11, distplot is deprecated

distplot

• Consolidate the original code to generate the distplot

for year in df.RY.unique():
    values = df.Value[df.RY == year]
    sns.distplot(values, hist=False, rug=True)

facetgrid

• properly configure the mapping and add hue to FacetGrid

g = sns.FacetGrid(df, col='MAJ_CAT', hue='RY')
p1 = g.map(sns.distplot, 'Value', hist=False, rug=True).add_legend()

Answered By - Trenton McKinney