I am working on an NLP project analyzing the words spoken by characters in The Office. Part of this project involves making a network diagram of which characters talk to each other for a given episode.
This will be shown in a Dash app by allowing a user to select dropdowns for 4 parameters: season, episode, character1, and character2.
Here is a relevant snippet of my code so far:
#Import libraries
import pandas as pd
import numpy as np
import dash
import dash_core_components as dcc
import dash_html_components as html
import dash_bootstrap_components as dbc
from dash.dependencies import Input, Output, State
#Load data
sheet_url = 'https://docs.google.com/spreadsheets/d/18wS5AAwOh8QO95RwHLS95POmSNKA2jjzdt0phrxeAE0/edit#gid=747974534'
url = sheet_url.replace('/edit#gid=', '/export?format=csv&gid=')
df = pd.read_csv(url)
#Set parameters
choose_season = df['season'].unique()
choose_episode = df['episode'].unique()
choose_character = ['Andy','Angela', 'Darryl', 'Dwight', 'Jan', 'Jim','Kelly','Kevin','Meredith','Michael','Oscar','Pam','Phyllis','Roy','Ryan','Stanley','Toby']
#Define app layout
app = dash.Dash()
server = app.server
app.layout = html.Div([
dbc.Row([
dbc.Col(
dcc.Dropdown(
id='dropdown1',
options=[{'label': i, 'value': i} for i in choose_season],
value=choose_season[0]
), width=3
),
dbc.Col(
dcc.Dropdown(
id='dropdown2',
options=[{'label': i, 'value': i} for i in choose_episode],
value=choose_episode[0]
), width=3
),
dbc.Col(
dcc.Dropdown(
id='dropdown3',
options=[{'label': i, 'value': i} for i in choose_character],
value=choose_character[0]
), width=3
),
dbc.Col(
dcc.Dropdown(
id='dropdown4',
options=[{'label': i, 'value': i} for i in choose_character],
value=choose_character[1]
), width=3
)
])
])
if __name__=='__main__':
app.run_server()
In order to have this work efficiently, I would like to have the following dependencies in the dropdown menus:
1.) The selection of the first dropdown menu updates the dropdown menu
ie: Season updates possible episodes
2.) The selection of the first two dropdown menus updates the 3rd and 4th dropdown menus
ie: Season, Episode updates possible characters (if a character was not in that episode, they will not appear)
3.) The selection of the third dropdown menu updates the fourth dropdown menu
ie: If a character is selected in the third dropdown menu, they can not be selected in the fourth (can't select the same character twice)
I understand one way to do this is to make a massive season to episode dictionary and then an even larger season to episode to character dictionary.
I've already made the code to process the season to episode dictionary:
#app.callback(
Output('dropdown2', 'options'), #--> filter episodes
Output('dropdown2', 'value'),
Input('dropdown1', 'value') #--> choose season
)
def set_episode_options(selected_season):
return [{'label': i, 'value': i} for i in season_episode_dict[selected_season]], season_episode_dict[selected_season][0]
I can definitely build these dictionaries, but this seems like a really inefficient use of time. Does anyone know of a way to build these dictionaries with just a few lines of code? Not sure how to approach building these in the easiest way possible. Also, if you have an idea for a better way to approach this problem, please let me know that too.
Any help would be appreciated! Thank you!
I think I see what you're asking about now. Something like this should get you a basic dictionary, which you could then modify for the options param for the dropdowns.
df = pd.read_csv(url)
season_episode_character_dictionary = {}
for season in df['season'].unique.tolist():
df_season = df[df['season'].eq(season)]
season_episode_character_dictionary[season] = {}
for episode in df_season['episode'].unique.tolist():
df_episode = df_season[df_season['episode'].eq(episode)]
characters = df_episode['characters'].unique.tolist()
season_episode_character_dictionary[season][episode] = characters
Related
I want to create a table in Dash where it is possible to choose multiple values in a specific column. My goal is to do it with a regular table and then add dropdowns for filtering.
However, when a dropdown is added and the size of the choices is bigger than the dropdown it adds "rows". Is it possible to let the choices that been made to be translated into number of choices made instead?
I wonder is it is possible to do something similiar to the Basic example in
https://mdbootstrap.com/docs/standard/extended/multiselect/
I have tried something like this
from dash import Dash, dcc, html, Input, Output
from plotly.express import data
import pandas as pd
df = data.medals_long()
app = Dash(__name__)
app.layout = html.Div([
dcc.Dropdown(df.columns, id='pandas-dropdown-1', multi=True),
html.Div(id='pandas-output-container-1')
])
#app.callback(
Output('pandas-output-container-1', 'children'),
Output('pandas-dropdown-1', 'search_value'),
Input('pandas-dropdown-1', 'value')
)
def update_output(value):
return f'You have selected {value}', f'{len(value)} values have been chosen'
if __name__ == '__main__':
app.run_server(debug=True)
The following code reads a simple model (3 columns 50 rows) from a CSV file for editing in a table in my (larger) dash app. Editing a cell writes the whole table back to file as expected. However, reloading the page displays the table as it was originally loaded from file, thus losing any edits. Any clues about how to keep the edits between page reloads?
df_topic_list=pd.read_csv(model_file)
app.layout = html.Div([
dcc.Store(id='memory-output'),
html.Div([
dash_table.DataTable(df_topic_list.to_dict('records'),
id='memory-table',
columns=[{"name": i, "id": i} for i in df_topic_list.columns],editable=True
),
])
])
#app.callback(Output('memory-output', 'data'),
Input('memory-table', 'data'))
def on_data_set_table(data):
pd.DataFrame(data).to_csv(model_file,index=False)
return data
app.run_server(port=8052)
When you refresh page then it doesn't run all code again but it only sends again app.layout which it generated only once and which has original data from file. And when you update data in cell in table then it updates only values in browser (using JavaScript) but not in code app.layout.
But it has options to presist values in browser memory and it should use these values after reloading.
app.layout = html.Div([
dcc.Store(id='memory-output'),
html.Div([
dash_table.DataTable(
df_topic_list.to_dict('records'),
id='memory-table',
columns=[{"name": i, "id": i} for i in df_topic_list.columns],
editable=True,
persistence=True, # <---
persisted_props=["data"], # <---
)
])
])
It works for me but it seems some people had problem with this.
See issues: Dash table edited data not persisting · Issue #684 · plotly/dash-table
But I found other method to keep it.
I assign table to separated variable - ie. table - and in callback I replace table.data in this table.
from dash import Dash, Input, Output, callback
from dash import dcc, html, dash_table
import pandas as pd
model_file = 'data.csv'
df_topic_list = pd.read_csv(model_file)
app = Dash(__name__)
table = dash_table.DataTable(
df_topic_list.to_dict('records'),
id='memory-table',
columns=[{"name": i, "id": i} for i in df_topic_list.columns],
editable=True,
)
app.layout = html.Div([
dcc.Store(id='memory-output'),
html.Div([table])
])
#app.callback(
Output('memory-output', 'data'),
Input('memory-table', 'data')
)
def on_data_set_table(data):
pd.DataFrame(data).to_csv(model_file, index=False)
table.data = data # <--- replace data
return data
app.run_server(port=8052)
I'm pretty confused by what 'children' even does in Dash html. Why does it even exist? Why would you use it? I tried reading the documentation, but it hasn't been very helpful.
With reference to the block of code below:
What does the children property even do in the first line?
Couldn't you replace children with something like 'figure' for the plots?
Block of Code:
app.layout = html.Div(children=[
# TODO1: Add title to the dashboard
html.H1("Airline Dashboard by CK", style = {'text-align':'center'}),
# REVIEW2: Dropdown creation
# Create an outer division
html.Div([
# Add an division
html.Div([
# Create an division for adding dropdown helper text for report type
html.Div(
[
html.H2('Report Type:', style={'margin-right': '2em'}),
]
),
# TODO2: Add a dropdown
dcc.Dropdown(id = 'input-type',
options = [{'label':'Yearly Airline Performance Report', 'value': 'OPT1'},
{'label':'Yearly Average Flight Delay Statistics', 'value': 'OPT2'}],
multi = False,
placeholder = 'Select a Report Type',
style={'width': '80%', 'padding': '3px', 'font-size': '20px', 'text-align-last': 'center'}
)
# Place them next to each other using the division style
], style={'display': 'flex'}),
# Add next division
html.Div([
# Create an division for adding dropdown helper text for choosing year
html.Div(
[
html.H2('Choose Year:', style={'margin-right': '2em'})
]
),
dcc.Dropdown(id='input-year',
# Update dropdown values using list comphrehension
options=[{'label': i, 'value': i} for i in year_list],
placeholder="Select a year",
style={'width': '80%', 'padding': '3px', 'font-size': '20px', 'text-align-last': 'center'}),
# Place them next to each other using the division style
], style={'display': 'flex'}),
]),
# Add Computed graphs
# REVIEW3: Observe how we add an empty division and providing an id that will be updated during callback
html.Div([], id='plot1'),
html.Div([
html.Div([], id='plot2'),
html.Div([], id='plot3')
], style={'display': 'flex'}),
# TODO3: Add a division with two empty divisions inside. See above disvision for example.
html.Div([
html.Div([], id='plot4'),
html.Div([], id='plot5')
], style = {'display':'flex'})
])
# Callback function definition
# TODO4: Add 5 ouput components
#app.callback(
[Input(component_id='input-type', component_property='value'),
Input(component_id='input-year', component_property='value')],
# REVIEW4: Holding output state till user enters all the form information. In this case, it will be chart type and year
[Output("plot1", 'children'), Output("plot2", "children"),
Output("plot3", "children"), Output("plot4", "children"),
Output("plot5", "children")
])
From this page of the docs:
The children property is special. By convention, it's always the first attribute which means that you can omit it: html.H1(children='Hello Dash') is the same as html.H1('Hello Dash'). Also, it can contain a string, a number, a single component, or a list of components.
Some components, such as html.Div and html.P, accept a value for their children prop. Others, such as dcc.Graph or dcc.Dropdown do not, and require other props in order to function properly.
As #KarlKnechtel mentioned in his comment, when one component is a child of another, it represents a nesting of the first component inside the other. The following are analogous:
In Dash:
html.Div(
children=[
html.H1('This is some text'),
html.P('This is also some text'),
]
)
In HTML:
<div>
<h1>This is some text</h1>
<p>This is also some text</p>
</div>
I hope that answers your questions.
Edit:
Adding style after children to this html.Div would allow you to change the styles of the Div, which may impact the styles of the components nested inside it, but that is not the purpose of the children prop. As the docs mention, you can either explicitly set children= whatever, or you can pass the same value in first, with no explicit key word argument, and Dash will treat that as the children prop. Either way, behind the scenes, the component is still receiving a value for its children property.
The purpose of the children property is to allow users to nest components, just as we do in raw HTML. Without the children prop, it would not be possible to do things like group related items together by containing them in the same parent element (ex. putting navigation items inside a top navigation bar).
I have an application in which I want to update the dropdown values when the user inputs some text into the dropdown. So far, I managed to get the text from an Input text that the user inputs, but not from the dropdown. The dropdown keeps reseting after clicking in any part of the screen or on the button.
import dash
import dash_html_components as html
import dash_core_components as dcc
import dash_bootstrap_components as dbc
from dash.dependencies import Input, Output, State
app = dash.Dash(
external_stylesheets=[
dbc.themes.BOOTSTRAP])
app.layout = html.Div(
[
dcc.Input(id='helper'),
dcc.Dropdown(id="my-dynamic-dropdown", options=[]),
html.Br(),
html.Br(),
html.Br(),
html.Br(),
html.Br(),
html.Br(),
html.Br(),
dbc.Button('Add to Dropdown', id='dropdown_button', n_clicks=0, color="info",
className="mr-1")
],
)
options = []
#app.callback(
Output("my-dynamic-dropdown", "options"),
Input('dropdown_button', 'n_clicks'),
[State('my-dynamic-dropdown', 'options'),
# State('helper', 'value')
State('my-dynamic-dropdown', 'search_value')
]
)
def update_options(n_clicks,existing_options,helper_value):
if helper_value is None:
print("none value")
return options
else:
print("helper value ",helper_value)
options.append({'label': helper_value, 'value': helper_value})
return options
if __name__ == "__main__":
app.run_server(debug=True)
The dropdown allows a user to search through the existing options, but what you're trying to do is not what the dropdown was designed for. You would either need to write a custom component, use some combination of dropdown and input components, or find a different solution to whatever problem you're trying to solve.
I am trying to pass input value to another with Dash Plot.
The main page has a list of strategies :
First_page = html.Div([
html.Div([dcc.Dropdown(id='strategy', options=[{'label': i, 'value': i} for i in [strat1, start2]], multi=True)], value = [strat1]),
html.Div([dcc.Input(id='date_start', type='text', value='1990-01-01')]),
html.Button('Start Calculation', id='button'),
])
And my pages 1 and 2 are the following :
for i in range(2) :
layout = html.Div([
html.Div([dcc.Dropdown(id='strategy'+str(i), options=[{'label': i, 'value': i} for i in [strat1, start2]]])], value = strat+str(i)),
html.Div([dcc.Input(id='date_start'+str(i), type='text', value='1990-01-01')]),
html.Button('Start Calculation', id='button'+str(i)),
+ other options not listed in main page
])
The goal is the following : when the user is on the main page he can choose the list of strategies he want to see, as well as the starting date.
Then when he press the button, it should open pages for each strategy selected (In the example we have 2 strategies so I should have 2 pages : localhost:port/page-strat1 and localhost:port/page-strat2) where the graph shown on each page starts with the selected date.
I succeed to open the pages localhost:port/strat1 and/or the page localhost:port/strat2 depending on the list of selected strategies.
list_event = [Event('button', 'click')]
list_state = [State('strategy', 'value')]
#app.callback(Output('hidden-div', 'children'),events=list_event, state=list_state)
def create_callback_share(datestart, list_strategies_used):
for _strat in list_strategies_used:
time.sleep(1)
webbrowser.open('http://localhost:port/page-'+_strat)
But I do not succeed to send the selected date. I try the following callback :
for i in range(2) :
#app.callback(Output('date_start'+str(i), 'value'),inputs=[Input('date_start', 'value')])
def set_display_children(selected_element):
return "{}".format(selected_element)
But the page localhost:port/strat1 and localhost:port/strat2 allow open with the default value which is '1990-01-01' even if I enter another value in the main page.
Is it possible to send value from one page to the other ?
Use dcc.Store() as in the answer here:
Plotly Dash Share Callback Input in another page with dcc.Store
And here is the Dash documentation:
https://dash.plotly.com/dash-core-components/store