opensci

import pandas as pd
import plotly.express as px

# Read in the data for this cluster (these were generated by generate_pages.py)
CLUSTER = "opensci"
df_hub = pd.read_csv("opensci-by-hub.csv", parse_dates=["date"])
df_month = pd.read_csv("opensci-month-end.csv", parse_dates=["date"])

# These are for the little number boxes at the top
current_mau = int(df_month["unique_users"].iloc[-1]) if not df_month.empty else 0
peak_mau = int(df_month["unique_users"].max()) if not df_month.empty else 0

order = (
    df_hub.groupby("hub")["users"].mean()
    .sort_values(ascending=False).index.tolist()
)
fig = px.area(
    df_hub.sort_values("date"),
    x="date", y="users", color="hub",
    category_orders={"hub": order},
    title=f"Active users by hub on {CLUSTER}",
    labels={"users": "users"},
)
fig.show()

fig = px.bar(
    df_month,
    x=df_month["date"].dt.strftime("%Y-%m"),
    y="unique_users",
    title=f"Monthly unique users on {CLUSTER}",
    labels={"x": "month", "unique_users": "unique users"},
)
fig.show()