andon.py

# AUTOGENERATED! DO NOT EDIT! File to edit: ../../nbs/book/EventTracker/11_Andon.ipynb.

# %% auto 0
__all__ = ['df', 'farms', 'selected_farms', 'filter_df', 'generate_discrete_colors', 'get_discrete_colormap', 'get_summary',
           'get_event_metrics', 'get_algo_performance', 'get_pie', 'get_data', 'get_farms', 'kpi']

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 2
import streamlit as st
import pandas as pd
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 6
st.set_page_config(
    page_title="Andon",
    layout='wide'
)

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 11
def generate_discrete_colors(n):
    colors = px.colors.qualitative.Plotly
    return [colors[i % len(colors)] for i in range(n)]

generate_discrete_colors(5)

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 12
def get_discrete_colormap(classes):
    colors = generate_discrete_colors(len(classes))
    color_discrete_map = dict(zip(classes, colors))
    return color_discrete_map

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 13
def get_summary(df, col_a='Type of Alert', col_b='Ground Truth'):
    cal_df = df[[col_a, col_b]].groupby([col_a, col_b]).value_counts().reset_index()

    
    color_discrete_map = get_discrete_colormap(df[col_b].dropna().unique())
    fig = px.bar(cal_df, y=col_a, 
                x='count', 
                color=col_b, 
                color_discrete_map=color_discrete_map, 
                orientation='h', 
                title=f'Summary: {col_a}|{col_b}')

    fig.update_layout(
        margin=dict(l=10, r=10, t=40, b=40),
        yaxis=dict(automargin=True)
    )

    return fig

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 16
def get_event_metrics(validated_df, 
                      event_type_cols=['Heat', 'Missed Heat'], 
                      event_type='Heat'):
    val_event_df = validated_df[validated_df['Type of Alert'].isin(event_type_cols)]
    a = val_event_df['Ground Truth'].value_counts()
    
    recall  = a.get('TRUE', 0)*100/(a.get('TRUE', 0)+a.get('MISSED', np.finfo(float).eps))
    precision = a.get('TRUE', 0)*100/(a.get('TRUE', 0)+a.get('FALSE', np.finfo(float).eps))

    return {
        f'Validated_{event_type}_Recall': recall,
        f'Validated_{event_type}_Precision': precision,
    }

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 17
def get_algo_performance(df):
    confirmed_df = df[df['Ground Truth'].isin(['TRUE','FALSE', 'MISSED', 'EXCLUDE', 'OTHER'])]
    validated_df = df[df['Ground Truth'].isin(['TRUE','FALSE', 'MISSED'])]
    missed_df = df[df['Ground Truth'].isin(['MISSED'])]
    heat_metrics = get_event_metrics(validated_df, 
                        event_type_cols=['Heat', 'Missed Heat'], 
                        event_type='Heat')
    health_metrics = get_event_metrics(validated_df, 
                        event_type_cols=['Health', 'Missed Health'], 
                        event_type='Health')

    data_coverage_metrics = {'Total':len(df),
                             'Confirmed': len(confirmed_df),
                             'Validated': len(validated_df),
                             'Missed': len(missed_df),
                             'Confirmed_Percentage': len(confirmed_df)*100/len(df),
                             'Validated_Percentage': len(validated_df)*100/len(df),
                             'Val_On_Confirm_Percentage': len(validated_df)*100/len(confirmed_df),
                             'Missed_On_Val_Percentage': len(missed_df)*100/len(validated_df) }
    performance_metrics = {**data_coverage_metrics,**heat_metrics, **health_metrics}
    return performance_metrics

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 20
def get_pie(df, colname):
    df_dist = df[colname].value_counts().reset_index()
    fig = px.pie(df_dist, values='count', 
                names=colname, 
                title=f'{colname}',
                color = colname,
                color_discrete_map = get_discrete_colormap(df[colname].dropna().unique())
                )
    return fig

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 35
def get_data(url="https://docs.google.com/spreadsheets/d/1TbqmmSzXtY8DwolVo9rN7M1QdZxLG30nro2cqrvn2CA/export?format=csv&#gid=294763682"):
    df = pd.read_csv(url).dropna(how='all')
    df['Alert Date'] = df['Alert Date'].str.replace("//", "-")
    df = df.dropna(subset=['Alert Date'])
    df['Type of Alert'] = df['Type of Alert'].str.strip()
    return df
# colors

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 36
def get_farms(df):
    return df['Dairy Farm'].dropna().unique().tolist()

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 39
def kpi(df, ncols = 7):
    metrics = get_algo_performance(df)
    dcols = st.columns(ncols)
    for (i, (k, v)) in enumerate(metrics.items()):
        with dcols[i%ncols]:
            st.metric(k, np.round(v, 2))

# %% ../../nbs/book/EventTracker/11_Andon.ipynb 40
#|eval: false
df = get_data()
farms = get_farms(df)
selected_farms = st.sidebar.multiselect('Select farms', farms, max_selections=len(farms))
filter_df = df
if selected_farms: 
    str_farms = "|".join(selected_farms)
    st.write(f"Selected_farms: {str_farms}")
    filter_df = df[df['Dairy Farm'].isin(selected_farms)]
else:
    st.write("Selected_farms: All")
# st.write(farms)
kpi(filter_df,ncols=4)
st.plotly_chart(get_pie(filter_df, colname = 'Type of Alert'))
st.plotly_chart(get_pie(filter_df, colname = 'Ground Truth'))
st.plotly_chart(get_pie(filter_df, colname = 'Dairy Farm'))
st.plotly_chart(get_summary(filter_df, col_a='Type of Alert', col_b='Ground Truth'))
st.plotly_chart(get_summary(filter_df, col_b='Type of Alert', col_a='Dairy Farm'))
st.plotly_chart(get_summary(filter_df, col_b='Ground Truth', col_a='Dairy Farm'))
#