Search code examples
pythonpandasseaborn

I'm unable to specify hue with wide-form data

I am trying to plot kernel density estimation plots for a data set that I successfully melted from a previous question.

This is what I should be getting (This was created using pd.concat([pd.DataFrame[Knn], pd.DataFrame[Kss], pd.DataFrame[Ktt], ...):

enter image description here

Here is what the dataframe looks like:

df_CohBeh
Out[122]: 
    melt         value
0    Knn  2.506430e+07
1    Knn  3.344882e+06
2    Knn  5.783376e+07
3    Knn  1.687218e+06
4    Knn  2.975834e+06
..   ...           ...
106  Ktt  2.056249e+08
107  Ktt  2.085805e+08
108  Ktt  7.791227e+07
109  Ktt  2.072576e+08
110  Ktt  4.658559e+07

[111 rows x 2 columns]

Where the column melt is simply the variable defined to specify three categories.

# In[parameter distribution]

# Melt the results to create a single dataframe
df_CohBeh = pd.melt(df, value_vars=['Knn', 'Kss', 'Ktt'], var_name='melt')

# Normal distribution plots
f, ax = plt.subplots()
sns.set_context("paper", 
                rc={"font.size":12,"axes.titlesize":8,"axes.labelsize":12})

ax = sns.displot(data=df_CohBeh, hue=['Knn', 'Kss', 'Ktt'], 
                 kind="kde", fill=True, legend=False, height=5, aspect=1.6, 
                 cut=0, bw_adjust=1)

ax.set(xlabel='Cohesive Parameters [Pa]', ylabel='Kernel Density Estimation')

# Legend
plt.legend(labels=[r'$K_{nn}$', r'$K_{ss}$', r'$K_{tt}$'], 
           loc='best').set_title("Parameter")

Here is the associated error message when I include hue=['Knn', 'Kss', 'Ktt']

ValueError: The following variable cannot be assigned with wide-form data: `hue`

When I remove hue=['Knn', 'Kss', 'Ktt'] from the displot function call, here is the resulting plot. I'm not sure where the error is that I am getting that isn't plotting correctly. enter image description here

Any help would be appreciated. Thank you!

Solution

  • When you already have the data in long format, you can specify the x values using x= and the group or color with hue = , and you just specify the column name from the dataframe, for example:

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

df = pd.DataFrame(np.random.poisson(10,(20,3)),
                  columns=['Knn', 'Kss', 'Ktt'])

df_CohBeh = pd.melt(df, value_vars=['Knn', 'Kss', 'Ktt'], var_name='melt')

sns.displot(data=df_CohBeh, hue='melt',x='value',
            kind="kde", fill=True, legend=False, height=5, aspect=1.6, 
            cut=0, bw_adjust=1)

    enter image description here