Observa el ritmo estacional con un mapa de calor de calendario

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

date_index = pd.date_range("2024-01-01", "2024-12-31", freq="D")

# Metrica ficticia con estacionalidad + efecto semanal + ruido
seasonal = 40 + 25 * np.sin(2 * np.pi * date_index.dayofyear / 365)
weekday_boost = np.where(date_index.weekday < 5, 8, -10)
rng = np.random.default_rng(42)
metric = seasonal + weekday_boost + rng.normal(0, 5, len(date_index))

calendar_df = pd.DataFrame({"date": date_index, "value": metric})
calendar_df["month"] = calendar_df["date"].dt.month
calendar_df["weekday"] = calendar_df["date"].dt.weekday

month_start_weekday = (
    calendar_df["date"].dt.to_period("M").dt.to_timestamp().dt.weekday
)
calendar_df["week_of_month"] = (
    (month_start_weekday + calendar_df["date"].dt.day - 1) // 7
).astype(int)

fig, axes = plt.subplots(3, 4, figsize=(12, 7), sharex=True, sharey=True)
vmin, vmax = calendar_df["value"].min(), calendar_df["value"].max()

for month in range(1, 13):
    ax = axes[(month - 1) // 4, (month - 1) % 4]
    month_df = calendar_df[calendar_df["month"] == month]
    matrix = np.full((7, 6), np.nan)

    for row in month_df.itertuples():
        matrix[row.weekday, row.week_of_month] = row.value

    im = ax.imshow(
        matrix,
        origin="upper",
        aspect="auto",
        cmap="viridis",
        vmin=vmin,
        vmax=vmax,
    )
    ax.set_title(f"Mes {month}", fontsize=11, pad=8)
    ax.set_xticks(range(6), labels=["Semana 1", "Semana 2", "Semana 3", "Semana 4", "Semana 5", "Semana 6"], fontsize=8)
    ax.set_yticks(range(7), labels=["Lun", "Mar", "Mie", "Jue", "Vie", "Sab", "Dom"], fontsize=8)
    ax.tick_params(length=0)

fig.suptitle("2024 Duracion diaria de sesion (min)", fontsize=14, y=0.98)
cbar = fig.colorbar(im, ax=axes, orientation="horizontal", fraction=0.035, pad=0.08)
cbar.set_label("Duracion promedio (min)")

fig.tight_layout()

plt.show()