gistfile1.txt

import pandas as pd

url = 'http://atmenv.envi.osakafu-u.ac.jp/omu-content/uploads/sites/1215/2015/10/KyotoFullFlower7.xls'
cherry_raw = pd.read_excel(url, skiprows=25, dtype_backend='pyarrow')

def tweak_cherry(df):
    return (df
            #.query('~`Full-flowering date (DOY)`.isna()')
            .rename(columns={'AD': 'year',
                             'Full-flowering date (DOY)': 'flowering_doy',
                             'Full-flowering date': 'flowering_MDD'})
            .assign(
              month=lambda df_: df_['flowering_MDD']//100,
              day=lambda df_: df_['flowering_MDD'] - (df_['month'] *100),
              #date=lambda df_: pd.to_datetime(df_[['year', 'month', 'day']], format='%Y%m%d', errors='coerce')
            )
            #.drop(columns=['month', 'day'])
            #.dropna(subset=['date'])
            #.assign(rolling_doy=lambda x: x['flowering_doy'].rolling(window=20).mean())
            #.loc[:, ['year', 'date', 'flowering_doy', 'rolling_doy', 'flowering_MDD']]
           )



def plot_cherry(df):
    plt.rcParams["font.family"] = "Roboto"
    figsize = (160, 165) # pts
    def points_to_inches(points):
        return points / 72
    figsize_inches = [points_to_inches(dim) for dim in figsize]
    padding = 6 # pts
    heading_fontsize = 11
    heading_fontweight = 'bold'
    subheading_fontsize = 9.5
    subheading_fontweight = 'normal'
    source_fontsize = 6.5
    source_fontweight = 'light'
    axis_fontsize = 7.5
    axis_fontweight = 'normal'
    tick_fontsize = 7
    tick_fontweight = 'light'

    fig, ax = plt.subplots(figsize=figsize_inches, dpi=300)
    df.plot.scatter(x='year', y='flowering_doy', ax=ax, c='#e7afbe',
                    s=3.5)
    df.plot(x='year', y='rolling', ax=ax, c='#bd4758', linewidth=1.5)
    # add label color of line to end of line
    ax.text(df['year'].iloc[-1], df['rolling'].iloc[-1], '20-year\n average', color='#bd4758',
                            fontsize=tick_fontsize, fontweight='normal')
    # set title
    plt.figtext(0.0 , 1, 'Cherry trees have been blossoming earlier\ndue to warmer spring temperatures', ha='left', 
                fontsize=heading_fontsize, fontweight=heading_fontweight)
    # set subtitle
    plt.figtext(0, 0.93, 'Date of peak cherry tree (Prunus jamasakura) blossom in Kyoto, Japan',
                fontsize=subheading_fontsize, fontweight=subheading_fontweight)
    # get rid of the spines
    for spine in ax.spines.values():
        spine.set_visible(False)
    # get rid of y-axis ticks
    ax.tick_params(axis='y', which='both', left=False)
    # set font size and weight for x-axis labels
    for label in ax.get_xticklabels():
        label.set_fontsize(tick_fontsize)
        label.set_fontweight(tick_fontweight)
    for label in ax.get_yticklabels():
        label.set_fontsize(tick_fontsize)
        label.set_fontweight(tick_fontweight)
    # set ticks at location/labels
    # [(70, 'March 11'), (80, 'March 21'), (90, 'March 31'), (100, 'April 10'), (110, 'April 20'), (120, 'April 30')],
    ax.set_yticks([70, 80, 90, 100, 110, 120])
    ax.set_yticklabels(['March 11', 'March 21', 'March 31', 'April 10', 'April 20', 'April 30'])
    # add dashed grid to y-axis
    ax.grid(axis='y', linestyle='--', alpha=0.7)
    # set x ticks to 812, 1000, 1200, 1400, 1600, 1800, 2023
    ax.set_xticks([812, 1000, 1200, 1400, 1600, 1800, 2023])
    # get rid of x and y axis labels
    ax.set_xlabel('')
    ax.set_ylabel('')
    # add source text at the bottom
    plt.figtext(0.00, -0.10, 'Source: Yasuki Aono (2021; 2024)', ha='left', 
                fontsize=source_fontsize, fontweight=source_fontweight,
                color='gray')
    # hide legend
    ax.legend().set_visible(False)
    plt.show()

(tweak_cherry(cherry_raw)
 .assign(rolling=lambda df_: df_['flowering_doy'].interpolate().rolling(window=20).mean())
 .pipe(plot_cherry)
)