We Keep Coding

How to update the data plot in a Bokeh app

I am having a Pandas DataFrame with a TimeSeries Index. The idea is to use the bokeh server in order to have a DateSlider. This means the values that I pass over the function are dates. However, I think my problem is the data is not updated when I am running the bokeh server. The Slider appears but the data is not updating while moving the cursor from the Slider. I would like to have your view. This is my code:
#The file was already imported and cleaned as DF: trt_file
day_one = str(min(trt_file.index)) #Checking point file
day_last = str(max(trt_file.index)) #Checking point file
# Creating the data source which will be selected from the whole file
source = ColumnDataSource(data={
"x" : trt_file.loc[day_one]["Depth [m]"],
"y" : trt_file.loc[day_one]["Temperature"],
"Temperature" : trt_file.loc[day_one]["Temperature"],
"Depth" : trt_file.loc[day_one]["Depth [m]"]
})
#Minimum values for axis generation
xmin, xmax = min(trt_file["Depth [m]"]), max(trt_file["Depth [m]"])
ymin, ymax = min(trt_file["Temperature"]), max(trt_file["Temperature"])
#Set the figure fopr the initial draw
p = figure(title="2018-07-09", x_axis_label="Depth [m]", y_axis_label="Temperature", plot_height=300, plot_width=1500, x_range=(xmin, xmax), y_range=(ymin, ymax),tools=[HoverTool(tooltips=[("Temperature", "#Temperature"), ("Depth [m]", "#Depth")])] )
#Draw the values from the source we defined first
p.circle(x="x", y="y", fill_alpha=0.5, source=source)
#Set the axis names
p.xaxis.axis_label = "Kable lenght [m]"
p.yaxis.axis_label = "Temperature [C]"
#Creating the DateTime for the Slider
day_one_dt = (min(trt_file.index)) #Checking point file
day_last_dt = (max(trt_file.index))
#Slider
slider = DateSlider(start=day_one, end=day_last, value=day_one, step=1, title="TRT Day")
#Creating the bokeh interactive figure
def update_plot(attr, old, new): #attribute to change, old_val, new_val
#set day for the slider value
day_start = (str(slider.value)) #Value from the slider
new_data = {
"x" : trt_file.iloc[day_start]["Depth [m]"],
"y" : trt_file.iloc[day_start]["Temperature"],
"Temperature": trt_file.iloc[day_start]["Temperature"],
"Depth" : trt_file.iloc[day_start]["Depth [m]"]
}
source.trt_file = new_data
p.title.text = "Day TRT"
slider.on_change("value", update_plot)
layout = row(widgetbox(slider), p)
curdoc().title = "Temp"
curdoc().add_root(layout)

Bokeh knows nothing at all about a "trt file" so setting source.trt_file is not something Bokeh will notice or do anything with. The property Bokeh knows about and responds to is data:
source.data = new_data

Unable to update datasource in Bokeh python

I am using Bokeh 1.0.1. I am unable to update the data source in the Update method i.e src.data.update(new_src.data) doesn't seem to work. Below is the full code.
def modify_doc(doc):
def create_dataset(df, resample='D'):
# Resample the data
src = df.resample(resample).mean()
# Reset index for hovering
src.reset_index(inplace=True)
return ColumnDataSource(src)
def create_plot(src):
# Blank plot with correct labels
p = figure(plot_width=700, plot_height=300, x_axis_type="datetime",
title = 'Variation of Pollution',
x_axis_label = 'Time', y_axis_label = 'Pollution (µg/m³)')
p.line(source=src, x='Date & Time', y='pm2.5', line_width=2,
color='firebrick', line_alpha=0.5, legend='Actual')
hover = HoverTool(tooltips=[('Pollution', '#{pm2.5} µg/m³'),
('Air Temp', '#{Air Temp.} °C'),
('Temp', '(#{Min. Temp.}{0.2f}, #{Max. Temp.}{0.2f}) °C'),
('Dew Pt.', '#{Dew Pt.} °C'),
('Rainfall', '#Rainfall mm'),
('Wind Dir.', '#{Wind Dir.} °'),
('Wind Speed', '#{Wind Speed} km/hr'),
('Relative humidity', '(#{Min. RH}{0.2f}, #{Max. RH}{0.2f}) %')],
mode='vline')
p.add_tools(hover)
p.legend.click_policy = 'hide'
return p
# Update function takes three default parameters
def update(attr, old, new):
# Resampling list
re_list = ['D', 'W', 'M', 'A']
# Make a new dataset based on the selected carriers and the
# make_dataset function defined earlier
new_src = create_dataset(df,
resample = re_list[resample_button_group.active])
# Update the source used the quad glpyhs
src.data.update(new_src.data)
resample_button_group = RadioButtonGroup(labels=["Day", "Week", "Month", "Year"], active=1)
resample_button_group.on_change('active', update)
controls = WidgetBox(resample_button_group)
# Initial Plot
src = create_dataset(df)
p = create_plot(src.data)
layout = row(controls, p)
doc.add_root(layout)
# Set up an application
handler = FunctionHandler(modify_doc)
app = Application(handler)

You should be able to update the line glyph directly.
First, modify your plotting code to assign a name to the line glyph:
pm_line = p.line(
source=src,
x='Date & Time',
y='pm2.5',
line_width=2,
color='firebrick',
line_alpha=0.5,
legend='Actual',
name='pm_line' # Add this!
)
Then in your update function, replace your existing update line with the following:
pm_line = p.select_one({'name':'pm_line'})
pm_line.data_source.data = new_src.data

Create time series graph with rangeslider

I'm trying to create slider that as you drag the slider, the portion of the graph that is shown is only what is on the slider. For example, if you look at my graph below, if the slider was set to 1990, you would only see the lines from 1990 to 2016. I found a working example with plotly but I wanted to see if it could be done with Bokeh.
This is my code so far:
p = figure(width = 900, height = 450)
p.xaxis.axis_label = 'Year'
p.yaxis.axis_label = 'Aggregated Number of Degrees in Education'
source = ColumnDataSource(df)
fill_source = ColumnDataSource(data=dict(x=[],y=[]))
# Create objects for each line that will be plotted
stem = p.line('year', 'stem', line_color='#8dd3c7', line_width=3, source=source)
stem = p.circle('year', 'stem', line_color='#8dd3c7', line_width=3, source=source)
sped = p.line('year', 'sped', line_color='#fdb462', line_width=3, source=source)
elem = p.line('year', 'elem', line_color='#bebada', line_width=3, source=source)
elem = p.square('year', 'elem', line_color='#bebada', line_width=3, source=source)
other = p.line('year', 'other', line_color='#fb8072', line_width=4, source=source)
aggtotal = p.line('year', 'aggtotal', line_dash=[4,4,], line_color='#80b1d3', line_width=3, source=source)
yaxis = p.select(dict(type=Axis, layout="left"))[0]
yaxis.formatter.use_scientific = False
legend = Legend(items=[("STEM", [stem])
,("SPED" , [sped])
,("Elementary", [elem])
,("Other", [other])
,("Total Education Graduates", [aggtotal])], location=(0, 0))
p.add_tools(HoverTool(tooltips=[("Date", "#year")]))
p.add_layout(legend, 'right')
callback_test = CustomJS(args=dict(source=source,fill_source=fill_source), code="""
var data = source.data;
var fill_data = fill_source.data;
var s_val = cb_obj.value;
fill_data['x']=[];
fill_data['y']=[];
for (i = 0; i < s_val; i++) {
fill_data['y'][i].push(data['y'][i]);
fill_data['x'][i].push(data['x'][i]);
}
fill_source.trigger('change');
""")
sped_slider = Slider(start=1984, end= 2016, value=1, step=1,title="Year",callback=callback_test)
callback_test.args["sped"] = sped_slider
layout = row(p,widgetbox(sped_slider))
This renders a slider but it doesn't do anything and I'm not sure where to go from here.

There are some issues with your callback code. For example:
you loop i from 0 to s_val (which may be 1990) which is not consistent with the length of your arrays.
your glyphs references columns 'stem', etc... but the fill_source has columns 'x' and 'y'
your glyphs reference source as a source but you change and trigger event on fill_source.
All that could probably be fixed but there's a much easier way, adjust the range in the callback. E.g. replace your callback by this:
x_range = p.x_range
callback_test = CustomJS(args=dict(x_range=x_range), code="""
var start = cb_obj.value;
x_range.start = start;
x_range.change.emit();
""")
Note the change to the event trigger. Your version would work but I think it's going to be deprecated.
Also:
this line callback_test.args["sped"] = sped_slider is not necessary
you could add toolbar_location='above' in figure(...) to avoid rendering conflict with the legend
you're still going to have a layout problem between the slider and the legend which can be fixed in different ways (slider under or put the slider and the legend in a column before adding to the right of the plot, etc...)

Bokeh Plot Update Using Slider

I am trying to use a slider to update my Bokeh Plot. I am finding it difficult to achieve it using pandas dataframe(did not find any examples so far).
The other way is to use the "columndatasource" (found some examples over forums) but still not able to achieve the functionality.
So I have two columns, X axis is date and the Y axis is Volume. I want to change my Y values based on slider input. I am able to see the plot but the slider functionality is not working
Any help will be very much appreciable.
source = ColumnDataSource(data=dict(x=df2['Date'],y=df2['Vol']))
S1 = figure(plot_width=400,plot_height=400,tools=TOOLS1,title="Volume Per Day",x_axis_type="datetime")
S1.line('x','y',source=source)
callback_test = CustomJS(args=dict(source=source), code="""
var data = source.get('data');
var s_val = cb_obj.value
x = data['x']
y = data['y']
console.log(cb_obj)
for (i = 0; i < s_val; i++) {
y[i] = y[i]
}
source.trigger('change');
""")
slider = Slider(start=0, end= max_Vol, value=1, step=100,title="Vol Per Day",callback=callback_test)

You are trying to update the range of data that is plotted using a slider.
When you do:
y = data['y']
for (i = 0; i < s_val; i++) {
y[i] = y[i]
}
the python equivalent would be, if y is some array with length>s_val:
for i in range(s_val):
y[i] = y[i]
This just replaces the elements from 0 to s_val-1 by themselves and doesn't change the rest of the list.
You can do two things:
update the displayed axis range directly
use an empty source that you will fill from your existing source based on the slider value
.
source = ColumnDataSource(data=dict(x=df2['Date'],y=df2['Vol']))
fill_source = ColumnDataSource(data=dict(x=[],y=[]))
S1 = figure(plot_width=400,plot_height=400,tools=TOOLS1,title="Volume Per Day",x_axis_type="datetime")
S1.line('x','y',source=fill_source)
callback_test = CustomJS(args=dict(source=source,fill_source=fill_source), code="""
var data = source.data;
var fill_data = fill_source.data;
var s_val = cb_obj.value;
fill_data['x']=[];
fill_data['y']=[];
for (i = 0; i < s_val; i++) {
fill_data['y'][i].push(data['y'][i]);
fill_data['x'][i].push(data['x'][i]);
}
fill_source.trigger('change');
""")

Here is the changes I have made to make it work with Bokeh last version
Some syntax error in the JavaScript part have been corrected, the method to trigger change is now change.emit, and the callback for a stand alone document is set after the Slider definition thanks to js_on_change
I have added all the import commands to give a complete example
I have also changed the visualization to show only data below the number of flight set by the slider (for more comprehension when moving the Slider towards lower values)
Below is the resulting code:
from bokeh.layouts import column, widgetbox
from bokeh.models import ColumnDataSource, CustomJS
from bokeh.models.widgets import Slider
from bokeh.plotting import Figure
import pandas as pd
from datetime import datetime, date, timedelta
from bokeh.plotting import show
from random import randint
today = date.today()
random_data = [[today + timedelta(days = i), randint(0, 10000)] for i in range(10)]
df2 = pd.DataFrame(random_data, columns = ['Date', 'Vol'])
source = ColumnDataSource(data = dict(x = df2['Date'], y = df2['Vol']))
fill_source = ColumnDataSource(data = dict(x = df2['Date'], y = df2['Vol'])) # set the graph to show all data at loading
TOOLS1 = []
S1 = Figure(plot_width = 400, plot_height = 400, tools = TOOLS1, title = "Volume Per Day", x_axis_type = "datetime")
S1.line('x', 'y', source = fill_source)
callback_test = CustomJS(args = dict(source = source, fill_source = fill_source), code = """
var data = source.data;
var fill_data = fill_source.data;
var s_val = cb_obj.value;
fill_data['x']=[];
fill_data['y']=[];
for (var i = 0; i <= data.x.length; i++) { // added "var" declaration of variable "i"
if (data['y'][i] <= s_val) { // more meaningful visualization: assuming you want to focuss on dates with less number of flights
fill_data['y'].push(data['y'][i]); // [i] index on left side of assignment removed
}
else {
fill_data['y'].push(0);
}
fill_data['x'].push(data['x'][i]);
}
fill_source.change.emit() ; // "trigger" method replaced by "change.emit"
""")
max_Vol = df2['Vol'].max()
slider = Slider(start = 0, end = max_Vol, value = max_Vol, step = 100, title = "Vol Per Day") # Remove attribute "callback = callback_test"
slider.js_on_change('value', callback_test) # new way of defining event listener
controls = widgetbox(slider)
layout = column(controls, S1)
show(layout)
Would be nice if I could embbed the resulting (HTML) visualization directly in this answer, let me now if it's possible ;)

Dynamic periodic update python bokeh

I have the following bokeh code how do you add a horizontal line or glyph to the chart dynamically do I need a callback?
In the documentation http://docs.bokeh.org/en/latest/docs/user_guide/interaction/callbacks.html it says "Custom callbacks like these can be set using a CustomJS object and passing it as the callback argument to a Widget object."
So do I set the call back before or after the update...
source = ColumnDataSource(dict(
time=[], average=[], low=[], high=[], open=[], close=[],
ma=[], macd=[], macd9=[], macdh=[], color=[]
))
#main chart
p = figure(plot_height=500, tools="xpan,xwheel_zoom,xbox_zoom,reset,crosshair,hover", x_axis_type=None, y_axis_location="right")
p.x_range.follow = "end"
p.x_range.follow_interval = 100
p.x_range.range_padding = 0
p.axis.minor_tick_in = -2
p.axis.minor_tick_out = 5
p.segment(x0='time', y0='low', x1='time', y1='high', line_width=2, color='black', source=source)
p.segment(x0='time', y0='open', x1='time', y1='close', line_width=8, color='color', source=source)
...
#count()
def update(t):
open, high, low, close, average = _get_prices(t)
color = "green" if open < close else "red"
new_data = dict(
time=[t],
open=[open],
high=[high],
low=[low],
close=[close],
color=[color],
)
print(open)
source.stream(new_data, 300)
output_notebook()
curdoc().add_periodic_callback(update, 50)
curdoc().title = "OHLC"