Search code examples
pythonmatplotlibseaborn

Formatting datetime64 dates as xticks

In a DataFrame I have column "Datum2" with datetime64 formatted dates (YYYYMMDD) that I want to use as xtick labels. My current plot looks liket this:

enter image description here

I want to change the labels to show the year and the abbreviations of the months as shown (e.g. 2022, Nov, Dec, 2023, Jan, Feb etc).

Currently I have this code:

import seaborn as sns
import matplotlib.pyplot as plt
import scipy.stats as stats
import numpy as np

df_tagesspiegel_final['Datum2'] = pd.to_datetime(df_tagesspiegel_final['Datum'], format='%Y%m%d')

# Scatterplot erstellen
sns.scatterplot(x='Datum2', y = 'compound', data=df_tagesspiegel_final)

#Quintile als xticks speichern
xticks = [df_tagesspiegel_final['Datum2'].min(), df_tagesspiegel_final['Datum'].median(numeric_only=True), df_tagesspiegel_final['Datum'].max()]

plt.gca().set(xticks=xticks, xlabel='Datum', ylabel='compound', title='Compound-Sentiment im Zeitverlauf')

plt.show()

How would I go about formatting the dates accordingly?
I feel like this might be a start https://matplotlib.org/stable/api/dates_api.html#matplotlib.dates.ConciseDateConverter
but to be honest, I am very new to python and I'm in way over my head

Solution

  • Here is an example using matplotlib.dates for formatting a custom datetime-formatted x-axis using seaborn

    • Every January tick is replaced with the (new) year label instead of month
    • ConciseDateFormatter is used for auto-abbreviating the months

    E.g.,:

    import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import matplotlib.dates as mdates

# Sample data
df_tagesspiegel_final = pd.DataFrame(
    {
        "Datum": [
            "20220101",
            "20220115",
            "20220221",
            "20220610",
            "20220903",
            "20221016",
            "20230201",
            "20230215",
        ],
        "compound": [0.11, 0.21, 0.3, 0.25, 0.47, 0.32, 0.48, 0.5],
    }
)

df_tagesspiegel_final["Datum2"] = pd.to_datetime(
    df_tagesspiegel_final["Datum"], format="%Y%m%d"
)

# Scatterplot construction
fig, ax = plt.subplots(figsize=(6, 4), dpi=150)
sns.scatterplot(
    x="Datum2", y="compound", data=df_tagesspiegel_final, ax=ax, zorder=2
)

# Set the locators for the x-axis
months = mdates.MonthLocator()  # Every month
years = mdates.YearLocator()  # Every year

# Get current axes ("gca")
ax = plt.gca()

ax.xaxis.set_major_locator(months)
ax.xaxis.set_minor_locator(years)

# Set the date format
ax.xaxis.set_major_formatter(mdates.ConciseDateFormatter(months))

# Display the plot
plt.grid(
    True, which="both", linestyle="--", linewidth=0.5, zorder=1, alpha=0.5
)
plt.xlabel("Datum")
plt.ylabel("compound")
plt.title("Compound-Sentiment im Zeitverlauf")
plt.show()

# Show data structure of sample data
print(df_tagesspiegel_final)
print(df_tagesspiegel_final.info())

    gives:

    Matplotlib.dates datetime-formatted custom x-axis tick labels in seaborn scatter plot.

          Datum  compound     Datum2
0  20220101      0.11 2022-01-01
1  20220115      0.21 2022-01-15
2  20220221      0.30 2022-02-21
3  20220610      0.25 2022-06-10
4  20220903      0.47 2022-09-03
5  20221016      0.32 2022-10-16
6  20230201      0.48 2023-02-01
7  20230215      0.50 2023-02-15

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 8 entries, 0 to 7
Data columns (total 3 columns):
 #   Column    Non-Null Count  Dtype         
---  ------    --------------  -----         
 0   Datum     8 non-null      object        
 1   compound  8 non-null      float64       
 2   Datum2    8 non-null      datetime64[ns]
dtypes: datetime64[ns](1), float64(1), object(1)
memory usage: 320.0+ bytes
None