Plot year over year data with month-date on X axis

March 7, 2024

I have a Pandas Dataframe containing daily data for a variable for 3 years. I need plot a line chart where the X axis is 1/1 to 12/31 (month and day only, no year) and has 3 lines, one for each year. How do I plot this in matplotlib?

        DateTime    price
10      2007-10-06  35756.0
11      2007-10-06  34747.0
12      2007-10-07  35748.0
13      2007-10-07  34743.0
14      2007-10-08  35740.0
... ... ...
1519    2009-10-29  29564.0
1520    2009-10-30  32035.0
1521    2009-10-30  29397.0
1522    2009-10-31  32003.0
1523    2009-10-31  29256.0

For the above data, I need X axis to be dates from 1/1 to 12/31 and 3 lines, one each for 2007, 2008 and 2009

>Solution :

1- using month-day strings for the X-axis

You can extract the year and month-day (with strftime), then pivot and plot:

import pandas as pd
import numpy as np

# create a dummy dataset
df = pd.DataFrame({'DateTime': pd.date_range('2020-01-01', '2022-12-31'),
                   'price': np.sin(np.arange(1096)/80)
                  })

# if needed, ensure we have datetime
df['DateTime'] = pd.to_datetime(df['DateTime'])

# create new columns with year and month-day
# pivot and plot
(df.assign(year=lambda d: d['DateTime'].dt.year,
           day=lambda d: d['DateTime'].dt.strftime('%m-%d')
          )
   .pivot(index='day', columns='year', values='price')
   .plot()
)

Output:

Variant with seaborn.lineplot:

import seaborn as sns
import matplotlib.ticker as ticker

ax = sns.lineplot(data=df, x=df['DateTime'].dt.strftime('%m-%d'), y='price',
                  hue=df['DateTime'].dt.year)

ax.xaxis.set_major_locator(ticker.LinearLocator(10))

Output:

Time series with the extracted year and month-day:

       DateTime     price  year    day
0    2020-01-01  0.000000  2020  01-01
1    2020-01-02  0.012500  2020  01-02
2    2020-01-03  0.024997  2020  01-03
3    2020-01-04  0.037491  2020  01-04
4    2020-01-05  0.049979  2020  01-05
...         ...       ...   ...    ...
1091 2022-12-27  0.877742  2022  12-27
1092 2022-12-28  0.883663  2022  12-28
1093 2022-12-29  0.889445  2022  12-29
1094 2022-12-30  0.895088  2022  12-30
1095 2022-12-31  0.900592  2022  12-31

[1096 rows x 4 columns]

2- using a real datetime X-axis with a common year, then changing the formatting

Another option that could be nicer would be to set up a common leap year (e.g. 2000) for plotting to benefit from the nice MonthLocator ticks, then to change the format to hide the year:

import matplotlib.dates as mdates

ax = (df
   .assign(year=lambda d: d['DateTime'].dt.year,
           day=lambda d: d['DateTime'].add(pd.DateOffset(year=2000))
          )
   .pivot(index='day', columns='year', values='price')
   .plot()
)

# add a tick for each month
ax.xaxis.set_major_locator(mdates.MonthLocator(interval=1))
# change the format to the month name
ax.xaxis.set_major_formatter(mdates.DateFormatter('%b'))

Output:

Same logic with seaborn:

import seaborn as sns
import matplotlib.dates as mdates

ax = sns.lineplot(data=df, x=df['DateTime'].add(pd.DateOffset(year=2000)),
                  y='price', hue=df['DateTime'].dt.year)

ax.xaxis.set_major_locator(mdates.MonthLocator(interval=1))
ax.xaxis.set_major_formatter(mdates.DateFormatter('%b'))

Output: