python 中 seaborn 行 plot 上的标记

Question

New here so putting hyperlinks. My dataframe looks like this.

 HR     ICULOS  SepsisLabel PatientID
100.3      1         0          1
117.0      2         0          1
103.9      3         0          1
104.7      4         0          1
102.0      5         0          1
88.1       6         0          1

Access the whole file here . What I wanted is to add a marker on the HR graph based on SepsisLabel (See the file). Eg, at ICULOS = 249, Sepsis Label changed from 0 to 1. I wanted to show that at this point on graph, sepsis label changed. I was able to calculate the position using this code:

mark = dummy.loc[dummy['SepsisLabel'] == 1, 'ICULOS'].iloc[0]
print("The ICULOS where SepsisLabel changes from 0 to 1 is:", mark)
Output: The ICULOS where SepsisLabel changes from 0 to 1 is: 249

I Plotted the graph using the code:

plt.figure(figsize=(15,6))

ax = plt.gca()

ax.set_title("Patient ID = 1")
ax.set_xlabel('ICULOS')
ax.set_ylabel('HR Readings')
sns.lineplot(ax=ax, 
             x="ICULOS", 
             y="HR", 
             data=dummy, 
             marker = '^', 
             markersize=5, 
             markeredgewidth=1, 
             markeredgecolor='black', 
             markevery=mark)

plt.show()

This is what I got: Graph . The marker was supposed to be on position 249 only. But it is also on position 0. Why is it happening? Can someone help me out?

Thanks.

Answer 1

Working with markevery can be tricky in this case, as it strongly depends on there being exactly one entry for each patient and each ICULOS .

Here is an alternative approach, using an explicit scatter plot to draw the marker:

import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
import numpy as np

df = pd.DataFrame({'HR': np.random.randn(200).cumsum() + 60,
                   'ICULOS': np.tile(np.arange(1, 101), 2),
                   'SepsisLabel': np.random.binomial(2, 0.05, 200),
                   'PatientID': np.repeat([1, 2], 100)})
for patient_id in [1, 2]:
    dummy = df[df['PatientID'] == patient_id]
    fig, ax = plt.subplots(figsize=(15, 6))
    ax.set_title(f"Patient ID = {patient_id}")
    ax.set_xlabel('ICULOS')
    ax.set_ylabel('HR Readings')
    sns.lineplot(ax=ax,
                 x="ICULOS",
                 y="HR",
                 data=dummy)
    x = dummy[dummy['SepsisLabel'] == 1]["ICULOS"].values[0]
    y = dummy[dummy['SepsisLabel'] == 1]["HR"].values[0]
    ax.scatter(x=x,
               y=y,
               marker='^',
               s=5,
               linewidth=1,
               edgecolor='black')
    ax.text(x, y, str(x) + '\n', ha='center', va='center', color='red')
    plt.show()

For your new question, here is an example how to convert the 'ICULOS' column to pandas dates. The example uses date 20210101 to correspond with ICULOS == 1 . You probably have a different starting date for each patient.

df_fb = pd.DataFrame()
df_fb['Y'] = df['HR']
df_fb['DS'] = pd.to_datetime('20210101') + pd.to_timedelta(df['ICULOS'] - 1, unit='D')