examples/example1.txt

"""
World Population Dataset

Monitoring city population (per country) with upper/lower bounds intervals
"""

# -- filter on arbitrarily selected countries
countries = df.sample(5)['country'].values
# countries = ["Japan", "Argentina", "Greece", "Thailand", "Peru",  "Saudi Arabia", "Jordan" ,"United States"];

# -- add a new column to df (mean or std)
df = df.groupby("country", as_index=False).apply(
    lambda d: d.assign(city_population_avg=d["population"].std().astype(int))
);
# -- measure members distance from LCL
df = df.groupby("country", as_index=False).apply(
    lambda d: d.assign(upper_bound=abs((d["city_population_avg"]*1.1).astype(int)+d["city_population_avg"]))
);
df = df.groupby("country", as_index=False).apply(
    lambda d: d.assign(lower_bound=abs((d["city_population_avg"]*1.1).astype(int)-d["city_population_avg"]))
);


for country in countries:

    _df = df[df['country']==country]
    b = px.bar(_df, x="city", y="population", facet_col="country", facet_col_wrap=4)

    s = px.line(
        _df, x="city", y="city_population_avg", facet_col="country", facet_col_wrap=4, 
        color="city_population_avg"
    ).update_traces(line_color="orange")
    b.add_traces(s.data)

    u = px.line(
        _df, x="city", y="upper_bound", facet_col="country", facet_col_wrap=4,
        color="upper_bound"
    ).update_traces(line_color="green")
    b.add_traces(u.data)

    l = px.line(
        _df, x="city", y="lower_bound", facet_col="country", facet_col_wrap=4,
        color="lower_bound"
    ).update_traces(line_color="red")
    b.add_traces(l.data)

    st.markdown(f"# {country}")
    col1, col2 = st.columns(2)
    col1.plotly_chart(b)
    col2.write(_df)

    below_control = _df[_df['population']<_df['lower_bound']].shape[0]
    msg = f"{round((below_control/_df.shape[0]), 2)*100}% of the cities are below the `lower_bound` of population control"
    st.markdown(f"> ### {msg}")