0

I am trying to perform drill down operation for various graphs in Dash Plotly , but I am not able to find any tutorial or documentation which provides.

I have tried some tutorials from YouTube and some read some documentations from google , but there are not a proper way of code which can actually fulfill my requirements.

The requirement is that when user clicks on a bar in bar chart then it should display another bar chart on the same place representing inner layer data. Thanks in advance!

chirag mehta
  • 118
  • 1
  • 2
  • 7

1 Answers1

4

Checkout this example of Drill Down in Dash with the help of callback_context. enter image description here

In this example I am showcasing just a single level drill down to keep it simple but with few modifications multi -level drill down can be achieved. There’s a back button for going back to the original figure. The back button is shown only on the level two of the drill down and hides on the original bottom level.

Code:

import dash
import dash_core_components as dcc
import dash_bootstrap_components as dbc
from dash.dependencies import Input, Output
import plotly.express as px
import pandas as pd

app = dash.Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])

# creating a dummy sales dataframe
product_sales = {'vendors':['VANS','VANS','VANS','VANS','NIKE','NIKE','NIKE','ADIDAS','ADIDAS','CONVERSE','CONVERSE','CONVERSE'],
                 'products': ['Tshirts','Sneakers','Caps','Clothing','Sports Outfit','Sneakers','Caps','Accessories','Bags','Sneakers','Accessories','Tshirts'],
                 'units sold': [2,15,3,8,37,13,7,4,12,7,8,2]
                 }
product_sales_df = pd.DataFrame(product_sales)

# all vendors sales pie chart
def sales_pie():
    df = product_sales_df.groupby('vendors').sum().reset_index()
    fig = px.pie(df, names='vendors',
                 values='units sold', hole=0.4)
    fig.update_layout(template='presentation', title='Sales distribution per Vendor')
    return fig

# creating app layout
app.layout = dbc.Container([
    dbc.Card([
            dbc.Button('', id='back-button', outline=True, size="sm",
                        className='mt-2 ml-2 col-1', style={'display': 'none'}),
            dbc.Row(
                dcc.Graph(
                        id='graph',
                        figure=sales_pie()
                    ), justify='center'
            )
    ], className='mt-3')
])

#Callback
@app.callback(
    Output('graph', 'figure'),
    Output('back-button', 'style'), #to hide/unhide the back button
    Input('graph', 'clickData'),    #for getting the vendor name from graph
    Input('back-button', 'n_clicks')
)
def drilldown(click_data,n_clicks):

    # using callback context to check which input was fired
    ctx = dash.callback_context
    trigger_id = ctx.triggered[0]["prop_id"].split(".")[0]

    if trigger_id == 'graph':

        # get vendor name from clickData
        if click_data is not None:
            vendor = click_data['points'][0]['label']

            if vendor in product_sales_df.vendors.unique():
                # creating df for clicked vendor
                vendor_sales_df = product_sales_df[product_sales_df['vendors'] == vendor]

                # generating product sales bar graph
                fig = px.bar(vendor_sales_df, x='products',
                             y='units sold', color='products')
                fig.update_layout(title='<b>{} product sales<b>'.format(vendor),
                                  showlegend=False, template='presentation')
                return fig, {'display':'block'}     #returning the fig and unhiding the back button

            else:
                return sales_pie(), {'display': 'none'}     #hiding the back button

    else:
        return sales_pie(), {'display':'none'}

if __name__ == '__main__':
    app.run_server(debug=True)

Also check out this thread for more info.

Atharva Katre
  • 457
  • 1
  • 6
  • 17