read_excel_benchmark_plotting.py

import hvplot
import pandas as pd
import polars as pl


def read_excel_plot():
    df = pl.read_csv("output/read_excel_results.csv")

    df = df.with_columns(
        df["Engine"]
        .map_elements(
            lambda engine: {
                f"Pandas {pd.__version__}": "#F9CFF2",
                "xlsx2csv": "#73bfb8",
                "openpyxl": "#26413c",
                "calamine (new)": "#0075FF",
            }.get(engine)
        )
        .alias("color")
    )

    plot = df.plot.bar(
        x="Engine",
        y="Time",
        color="color",
        ylabel="Time (seconds)",
        title="Reading Excel performance comparison w/ different engines",
    )

    plot = plot.redim.values(
        Engine=["calamine (new)", "openpyxl", "xlsx2csv", f"Pandas {pd.__version__}"]
    )

    return hvplot.save(plot, "output/read_excel_plot.html")


read_excel_plot()