Dash Plotly - How to make 2 conditions to plot a histogram?

Question

I'm looking for a way to graph a histogram for data respecting 2 Dropdown . I have to choose the value of firstcall and the value of the secondcall in order to plot the histogram. I don't find lot of literature on this subject I hope one of you have already face to this.

Please find an excel file with some data and the code I try bellow:

import dash
import dash_core_components as dcc
import dash_html_components as html
import plotly.graph_objs as go
import pandas as pd

df = pd.read_excel(
    "/Users/appelexcel.xlsx"
)

mgr_options = df["premierappel"].unique()
mgr_options_second = df["secondappel"].unique()

external_stylesheets = ['https://codepen.io/chriddyp/pen/bWLwgP.css']

app = dash.Dash(__name__, external_stylesheets=external_stylesheets)

colors = {
    'background': '#FDFFFF',
    'text': '#0A25DC'
}

app.layout = html.Div(style={'backgroundColor': colors['background']},children=[
    html.H1(children='Call',
     style={
            'textAlign': 'center',
            'color': colors['text']
        }
        ),

    html.Div(
        [
            dcc.Dropdown(
                id="premierappel",
                options=[{
                    'label': i,
                    'value': i
                } for i in mgr_options],
                value='All First Call'),
        ],
        style={'width': '25%',
               'display': 'inline-block'}),
    dcc.Graph(id='secondcallgraph'), 
 #The first plot just give the 2nd call

    html.Div(
        [
            dcc.Dropdown(
                id="secondappel",
                options=[{
                    'label': i,
                    'value': i
                } for i in mgr_options_second],
                value='All Second Call'),
        ],
        style={'width': '25%',
               'display': 'inline-block'}),
    dcc.Graph(id='thirdcallgraph'), # second figure
])


@app.callback(
    dash.dependencies.Output('secondcallgraph', 'figure'),
    [dash.dependencies.Input('premierappel', 'value')])
def update_graph(premierappel):
    if premierappel == "All First Call":
        df_plot = df.copy()
    else:
        df_plot = df[df['premierappel'] == premierappel]

    #func=(lambda x: round(100*x.count()/df_plot.shape[0] ,2))
    pv = pd.pivot_table(
        df_plot,
        index=['Age_1_2'],
        columns=['secondappel'],
        values=['frequency_1_2'],
        aggfunc=sum,
        fill_value=0)

    trace1 = go.Bar(x=pv.index, y=pv[('frequency_1_2', 'modification')], name='Modification')
    trace2 = go.Bar(x=pv.index, y=pv[('frequency_1_2', 'informations')], name='Informations')
    trace3 = go.Bar(x=pv.index, y=pv[('frequency_1_2', 'autres')], name='Autres')
    trace4 = go.Bar(x=pv.index, y=pv[('frequency_1_2', 'achat')], name='Achat')

    return {
        'data': [trace1, trace2, trace3, trace4],
        'layout':
        go.Layout(
            title='Appel 2 / {}'.format(premierappel),
            xaxis=dict(
                title='Days after 1st Call'),
            yaxis=dict(
                title='Count'),
            barmode='stack')
    }

Second graph (on third call)

My problem appears here, how can I tell him to take into account 2 conditions (one on first call and one on the second call qualification) ?

@app.callback(
    dash.dependencies.Output('thirdcallgraph', 'figure'),
    [dash.dependencies.Input('premierappel', 'value'), dash.dependencies.Input('secondappel', 'value')])
def update_graph(premierappel,secondappel):
    if premierappel & secondappel == "All Second Call":
        df_plot = df.copy()
    else:
        df_plot = df[(df['premierappel']==premierappel) & (df['secondappel']==secondappel)]

    #func=(lambda x: round(100*x.count()/df_plot.shape[0] ,2))
    pv = pd.pivot_table(
        df_plot,
        index=['Age_2_3'],
        columns=['troisiemeappel'],
        values=['frequency_2_3'],
        aggfunc=sum,
        fill_value=0)

    trace1 = go.Bar(x=pv.index, y=pv[('frequency_2_3', 'modification')], name='Modification')
    trace2 = go.Bar(x=pv.index, y=pv[('frequency_2_3', 'informations')], name='Informations')
    trace3 = go.Bar(x=pv.index, y=pv[('frequency_2_3', 'autres')], name='Autres')
    trace4 = go.Bar(x=pv.index, y=pv[('frequency_2_3', 'achat')], name='Achat')

    return {
        'data': [trace1, trace2, trace3, trace4],
        'layout':
        go.Layout(
            title='Appel 2 / {}'.format(secondappel),
            xaxis=dict(
                title='Days after 2nd Call'),
            yaxis=dict(
                title='Count'),
            barmode='stack')
    }


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

Thanks for your time !

Age_._. : time between calls.

Frequence : should be frequency but it's really random just to see how it's working.

Please find the data above. https://docs.google.com/spreadsheets/d/1u7E6GwJj1nsjOwIQIntcCWCKsczw36_iHPiEV2bjMcs/edit?usp=sharing

Answer 1

I can't wrap my head around what it is supposed mean. But here is what I can assume from the error messages I am getting:

File "pandas\_libs\index.pyx", line 672, inpandas._libs.index.BaseMultiIndexCodesEngine.get_loc

KeyError: ('frequency_2_3', 'modification')

That's it.

In your second callback you are assigning four traces. But the variable (pandas.Series()) pv will only hold a single column (during my debugging trials).

Assuming that the second callback is supposed to work as a filter: The solution might involve a If ... elif ... else block.