[FIXED] Matplotlib or seaborn or pandas : Plot with "deep" discretization

Issue

I have data in the following form :

pkgSpot   state     id_data_x     id_data_y  pkgSpot_y  delay_mean  delay_max  delay_min
0           1      Free  7.245899e+08  7.245899e+08        1.0           0.334572             292.0            -161.0
1           1  Occupied  7.245876e+08  7.245876e+08        1.0           2.865248             116.0            -162.0
2           2      Free  7.245884e+08  7.245884e+08        2.0           0.122951             294.0             -84.0
3           2  Occupied  7.245885e+08  7.245885e+08        2.0           1.344130             257.0            -279.0
4           3      Free  7.245909e+08  7.245909e+08        3.0          -2.931159             261.0            -196.0
5           3  Occupied  7.245894e+08  7.245894e+08        3.0           1.975265             246.0            -273.0
6           4      Free  7.245753e+08  7.245753e+08        4.0           0.889908             222.0            -235.0
7           4  Occupied  7.245729e+08  7.245729e+08        4.0           1.483180             180.0            -117.0
8          17      Free  7.245742e+08  7.245742e+08       17.0         -10.535714             160.0            -236.0
9          17  Occupied  7.245744e+08  7.245744e+08       17.0           7.473988             294.0            -258.0
10         18      Free  7.246035e+08  7.246036e+08       18.0          -9.374269             104.0            -160.0
11         18  Occupied  7.246025e+08  7.246025e+08       18.0           8.403315              88.0            -100.0
12         19      Free  7.245642e+08  7.245642e+08       19.0          -4.568548             220.0            -271.0
13         19  Occupied  7.245633e+08  7.245633e+08       19.0           4.474790             253.0            -262.0
14         26      Free  7.245383e+08  7.245383e+08       26.0          -0.480363             280.0            -300.0
15         26  Occupied  7.245365e+08  7.245366e+08       26.0         -10.149856             263.0            -298.0
16         27      Free  7.245861e+08  7.245861e+08       27.0          -3.831683             300.0            -258.0
17         27  Occupied  7.245864e+08  7.245864e+08       27.0           1.077670             300.0            -299.0
18         28      Free  7.245878e+08  7.245878e+08       28.0          -8.868201             221.0            -300.0
19         28  Occupied  7.245891e+08  7.245891e+08       28.0           6.633684             241.0            -220.0

and I would like to have, in one figure, a graph showing the mean, max and min delay discretized per pkgSpot and per state

Is there an easy way to achieve that with either pandas, seaborn or matplotlib ? I have played a little bit with the three libraries and with the pandas melt function but I could not find a way to do that 'easily'.

Thanks for your support

Solution

Here is an approach using filled areas to show the minimum, mean and maximum:

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from io import StringIO

data_str = '''pkgSpot state delay_mean delay_max delay_min
  1     Free   0.334572 292.0  -161.0
  1 Occupied   2.865248 116.0  -162.0
  2     Free   0.122951 294.0   -84.0
  2 Occupied   1.344130 257.0  -279.0
  3     Free  -2.931159 261.0  -196.0
  3 Occupied   1.975265 246.0  -273.0
  4     Free   0.889908 222.0  -235.0
  4 Occupied   1.483180 180.0  -117.0
 17     Free -10.535714 160.0  -236.0
 17 Occupied   7.473988 294.0  -258.0
 18     Free  -9.374269 104.0  -160.0
 18 Occupied   8.403315  88.0  -100.0
 19     Free  -4.568548 220.0  -271.0
 19 Occupied   4.474790 253.0  -262.0
 26     Free  -0.480363 280.0  -300.0
 26 Occupied -10.149856 263.0  -298.0
 27     Free  -3.831683 300.0  -258.0
 27 Occupied   1.077670 300.0  -299.0
 28     Free  -8.868201 221.0  -300.0
 28 Occupied   6.633684 241.0  -220.0'''

df = pd.read_csv(StringIO(data_str), delim_whitespace=True)

fig, ax = plt.subplots(figsize=(12, 4))
for state, color in zip(['Free', 'Occupied'], ['dodgerblue', 'crimson']):
     df_state = df[df['state'] == state]
     x = df_state['pkgSpot'].astype(str)
     ax.plot(x, df_state['delay_mean'], color=color)
     ax.fill_between(x, df_state['delay_min'], df_state['delay_max'], color=color, alpha=0.4, label=state)
ax.set_xlabel('pkgSpot')
ax.set_ylabel('delay (min, mean, max)')
ax.margins(x=0.02)
ax.legend(ncol=2, loc='lower center', bbox_to_anchor=[0.5, 1.01])
plt.tight_layout()
plt.show()

Another option uses errorbars:

fig, ax = plt.subplots(figsize=(12, 4))
for state, color, dodge in zip(['Free', 'Occupied'], ['dodgerblue', 'crimson'], [-0.2, 0.2]):
     df_state = df[df['state'] == state]
     x = np.arange(len(df_state)) + dodge
     yerr = [df_state['delay_mean'] - df_state['delay_min'], df_state['delay_max'] - df_state['delay_mean']]
     ax.errorbar(x, df_state['delay_mean'], yerr=yerr, color=color, ls=':', lw=2, capsize=10, capthick=2, label=state)

ax.set_xticks(np.arange(len(df_state)))
ax.set_xticklabels(df_state['pkgSpot'].astype(str))
ax.set_xlabel('pkgSpot')
ax.set_ylabel('delay (min, mean, max)')
ax.legend(ncol=2, loc='lower center', bbox_to_anchor=[0.5, 1.01])
plt.tight_layout()
plt.show()

Answered By - JohanC