PMs of GB via https://en.wikipedia.org/wiki/List_of_prime_ministers_of_the_United_Kingdom_by_length_of_tenure (inspired by https://twitter.com/Jonathan_Pryor/status/1583196889435865088)

leafy-green_1f96c.png

tenure_data.tsv

tenure_data_preproc.py

import pandas as pd

__all__ = [
    "count_total_days",
    "url",
    "preprocess_table",
    "df",
]

url = "https://en.wikipedia.org/wiki/List_of_prime_ministers_of_the_United_Kingdom_by_length_of_tenure"


def count_total_days(tenure: str) -> int:
    years, days = [int(n) for n in ["0", *tenure.split()] if n.isnumeric()][-2:]
    years_in_days = int(365.25 * years)
    return years_in_days + days


def preprocess_table(url) -> pd.DataFrame:
    df = (
        pd.read_html(url, match="Tenure")
        .pop()
        .drop(columns="Ref.")
        .rename({"Rank": "Longest"})
        .apply(lambda src: src.replace(r"\s\(.*", "", regex=True))
    )
    df.columns = (
        pd.Series(df.columns)
        .str.encode("ascii", "ignore")
        .str.decode("ascii")
        .str.replace(r"[\[\.].*", "", regex=True)
        .str.replace("for exit", "", regex=False)
        .str.replace("length", "", regex=False)
        .str.strip(" ")
    )
    df.Reason = df.Reason.astype("category")
    df["Days"] = df.Tenure.apply(count_total_days)
    df["Year"] = pd.to_datetime(df.Start, format="%Y")
    shortest = df.Rank.to_list()[::-1]
    df = (
        pd.concat((pd.Series(shortest, name="Shortest"), df), axis=1)
        .sort_values(by="Shortest")
        .reset_index(drop=True)
    )
    return df


df = preprocess_table(url)

tenure_figure.png

tenure_figure_crop.png

tenure_plot_utils.py

import matplotlib.pyplot as plt
import pandas as pd

from tenure_data_preproc import df

__all__ = [
    "fig_scale",
    "colour_by_party",
    "get_max_days_that_year",
    "get_min_days_that_year",
    "getcol",
    "round_up",
]

fig_scale = 2

plt.rcParams.update(
    {
        "text.usetex": True,
        "font.family": "Helvetica",
        "font.size": 18 * (1.3 * fig_scale),
    }
)


def colour_by_party(vals):
    a, b = vals
    if b < 0:
        colour = "w"
    else:
        year = a.year
        days = int(b)
        matched = df.query("Start == @year & Days == @days").squeeze()
        assert not matched.empty, "No result!"
        if "Labour" in matched.Party:
            colour = "red"
        elif "Conservative" in matched.Party:
            colour = "blue"
        elif "Liberal" in matched.Party:
            colour = "orange"
        else:
            colour = "black"
    return colour


def getcol(values):
    return values.reset_index().apply(colour_by_party, axis=1)


def get_max_days_that_year(t, zero=float("-inf")):
    days = df["Days"][df.Start == t.year]
    return zero if days.empty else days.max()


def get_min_days_that_year(t, zero=float("-inf")):
    days = df["Days"][df.Start == t.year]
    return zero if days.empty else days.min()


def round_up(val, nearest=1000):
    return int(nearest * ((val // nearest) + 1))


# def getcol(values, default):
#     colours = [default if val > 0 else "w" for val in values]
#     return pd.Categorical(colours)

tenure_viz.py

from pathlib import Path

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from numpy import linspace

from tenure_plot_utils import (
    fig_scale,
    get_max_days_that_year,
    get_min_days_that_year,
    getcol,
    round_up,
)

__all__ = ["plot_pms"]


def plot_pms(df) -> None:
    times = pd.date_range(
        df.Year.min(), end=df.Year.max() + pd.to_timedelta(366, unit="d"), freq="1y"
    )
    fig, ax = plt.subplots(1)
    fig_w, fig_h = (15.5, 10.5)
    fig.set_size_inches(fig_w * fig_scale, fig_h * fig_scale)
    pad = int(15 * fig_scale * 1.5)
    title = "$UK$ $Prime$ $Ministerial$ $tenure$"
    ax.set_title(title, pad=pad)
    plt.ylim(ymin=0, ymax=round_up(df.Days.max()))

    # url = "https://emojipedia-us.s3.dualstack.us-west-1.amazonaws.com/thumbs/120/whatsapp/326/leafy-green_1f96c.png"
    filename = "leafy-green_1f96c.png"
    lettuce_p = Path(__file__).parent / filename
    opacity = 1.0  # [1.0, 1.0, 1.0, 0.1]  #
    lettuce_img = plt.imread(lettuce_p) * opacity

    ax.set_xlabel("Date entered office", labelpad=pad)
    ax.set_ylabel("Tenure /days", labelpad=pad)
    fig.autofmt_xdate()
    ax.minorticks_on()
    ax.tick_params("both", length=20, width=2, which="major")
    ax.tick_params("x", length=10, width=1, which="minor")
    min_values = times.to_series().apply(get_min_days_that_year)
    max_values = times.to_series().apply(get_max_days_that_year)
    # Most has been 2 per year, so max and min will cover both

    days_agg = df.groupby("Start")["Days"].agg(["min", "max"])
    df.merge(days_agg, on="Start").rename({"min": "MinTenure", "max": "MaxTenure"})

    markersize = 50 * fig_scale**2.5
    plt.scatter(times, min_values, c=getcol(min_values), s=markersize)  # alpha=0.5,
    plt.scatter(times, max_values, c=getcol(max_values), s=markersize)  # alpha=0.5,

    x = linspace(df.Start.min(), 2000, num=4) / 2
    y = np.ones_like(x) * 1000

    ax_width = ax.get_window_extent().width + 500
    fig_width = fig.get_window_extent().width
    fig_height = fig.get_window_extent().height
    lettuce_size = ax_width / (fig_width * 40)
    lettuce_axs = [None for i in range(len(x))]

    for i in range(len(x)):
        loc = ax.transData.transform((x[i], y[i]))
        old_x, old_y = (
            loc[0] / fig_width - lettuce_size / 2,
            loc[1] / fig_height - lettuce_size / 2,
        )
        x_proportion = 0.815
        y_proportion = 0.21
        lvals = [
            x_proportion,
            y_proportion,
            lettuce_size,
            lettuce_size,
        ]
        lettuce_axs[i] = fig.add_axes(
            lvals,
            anchor="C",
        )
        lettuce_axs[i].imshow(lettuce_img)
        lettuce_axs[i].axis("off")

    xdata, ydata = 0, 0
    xdisplay, ydisplay = ax.transData.transform((xdata, ydata))
    annotate = True
    if annotate:
        ax.annotate(
            "45\ndays",
            xy=(0.95, 0.008),
            xycoords="axes fraction",
            fontsize=int(10 * (1.4 * fig_scale)),
            xytext=(-(int(50 * fig_scale)), int(26 * fig_scale)),
            textcoords="offset points",
            bbox=dict(boxstyle="round", ec="white", fc="white", alpha=0.0),
            arrowprops=dict(
                arrowstyle="-",
                connectionstyle="angle,angleA=120,angleB=0,rad=10",
            ),
        )

    legenditems = [
        (
            plt.Line2D(
                [],
                [],
                color="w",
                marker="o",
                markerfacecolor=c,
                # alpha=0.9,
                ms=markersize / 10,
            ),
            party,
        )
        for party, c in {
            "Labour": "red",
            "Conservative": "blue",
            "Liberal": "orange",
            "Other": "black",
        }.items()
    ]
    ax.legend(*zip(*legenditems), fontsize="small")
    show = False
    if show:
        plt.show()
    else:
        plt.savefig("tenure.png")

uk_pm_tenure.py

from tenure_viz import plot_pms

from tenure_data_preproc import df  # isort: skip

plot_pms(df)
# df.to_csv("tenure_data.tsv", index=False, sep="\t")