Dash DataTable - Unable to use more than 1 table

I have been using the DataTable from dash-table-experiments and it looks very cool.
I have been unable to have 2 (or more) tables appear on the same line, though.
Anyone knows if this is possible?

@pietroppeter - Hm, I’m not sure what’s going wrong here. Could you paste a simple reproducable example?

thanks for the reply, I will post a simple example and in the meantime I will try to explain better.

There is probably nothing wrong with dash here, only I am unable to control the width of the table in the final html (and I have just started using the tool, so it might be I am missing something very simple). I think it is more a problem of html/css (of which I am no expert at all) than of dash, which I am liking a lot!

I will like to have two tables one next to the other in the final html (right below the chart, they represent filters through different dimensions - product, market - of sales weekly data I want to plot). I tried to add some “display: inline-block” in container divs and using bootstrap css, but each of the data table ends up always taking the full width (I admit I also tried to tweak the dash_experiments_table.css, but no luck in that too).


I was able to solve my problem (I need to put also a width constraint in the div, not only the display:inline-block).
Now the two tables are one next to the other,as I wanted (see image).

Below is the reproducible code.

Using this code there is a weird behaviour (bug?) happening from the interaction of the 2 data tables.
In particular sometimes when clicking the “header checkbox” of the right table (the one that selects/deselects all), the click actually sets/unsets the header checkbox of the left table.
It is weird since I could not find a way to reproduce it all the time in the exact same way, but repeating the following steps I run into it sooner or later (usually in less than 5 iterations):

  • click on checkbox for product 1
  • click on checkbox for market 1
  • click on header checkbox for markets
  • if the bug happens stop (or you can repeatedly click on the market header checkbox and see the product header checkbox being set/unset), otherwise repeat the above steps.


import numpy as np
import pandas as pd
from itertools import product
import dash
import dash_html_components as html
import dash_core_components as dcc
import dash_table_experiments as dt
from dash.dependencies import Input, Output
import plotly.graph_objs as go


# ------------------------------------------------------------------------------------------------
# data generator
# multidimensional time series data
def base_signal(length, qty_start, qty_end, noise):
    """simple signal with trend and noise"""
    x = np.linspace(qty_start, qty_end, length)
    return x * np.random.normal(loc=1, scale=noise, size=length)

def seasonality(length, period, phase, amplitude):
    """multiplicative seasonal coefficient"""
    x = np.linspace(0, length * 2 * np.pi / period, length)
    x += phase * 2 * np.pi
    return 1 + amplitude * np.sin(x)

n_weeks = 52

# products
n_products = 20
df_product = pd.DataFrame({
    'product_id': range(n_products),
    'product_name': ['product %d' % i for i in range(n_products)],
    'qty_start': np.random.uniform(50, 150, n_products),
    'qty_end': np.random.uniform(50, 150, n_products),
    'noise': np.random.uniform(0, 0.2, n_products)
df_product['trend'] = (df_product['qty_end'] - df_product['qty_start']) / n_weeks
df_product = df_product['product_name trend noise qty_start qty_end'.split()]

# markets
n_markets = 20
df_market = pd.DataFrame({
    'market_id': range(n_markets),
    'market_name': ['market %d' % i for i in range(n_markets)],
    'period': np.random.choice([7, 13, 52], size=n_markets),
    'amplitude': np.random.uniform(0, 0.5, n_markets)
df_market['phase'] = df_market.period.apply(lambda z: np.random.choice([0.5*z, -0.5*z]))
df_market = df_market['market_name period amplitude phase'.split()]

# time series
df = pd.DataFrame(columns='product_id market_id week_id demand'.split())
for p, m in product(df_product.index, df_market.index):
    x = base_signal(
    df = df.append(pd.DataFrame({'product_id': p,
                                 'market_id': m,
                                 'week_id': range(n_weeks),
                                 'demand': x
df.set_index('product_id market_id week_id'.split(), inplace=True)

# ------------------------------------------------------------------------------------------------
# Dash app
# time series bar plot with two dimension filters

app = dash.Dash()

demand_graph = dcc.Graph(id='demand-graph')

filters = html.Div([
            selected_row_indices=list(df_product.index),  # all rows selected by default
    ], style={'width': '50%', 'display': 'inline-block'}),
            selected_row_indices=list(df_market.index),  # all rows selected by default
    ], style={'width': '50%', 'display': 'inline-block'})

app.layout = html.Div(children=[demand_graph, filters])

    Output('demand-graph', 'figure'),
    [Input('product-datatable', 'selected_row_indices'),
     Input('market-datatable', 'selected_row_indices')])
def update_figure(selected_products, selected_markets):
    # filter and group
    dff = df.loc[tuple([selected_products, selected_markets, slice(None)])]
    df_chart = dff.groupby('week_id').sum()

    data = [go.Bar(x=df_chart.index, y=df_chart['demand'], name='demand')]

    title = 'Demand data for %d products and %d markets' % (len(selected_products), len(selected_markets))
    layout = go.Layout(title=title)

    return {'data': data, 'layout': layout}

if __name__ == '__main__':

@pietroppeter @chriddyp
I am also facing same problem, did you get any help.
I have also used data table two time in a page , the second table operations is overlapping with first, i am waiting for my luck.