I am wondering if there's an easy way to build an iPyWidget with a datetime slider. Right now it is easy to slide over integer or floating point ranges (e.g. numbers 1-10, decimals 0.01, 0.02, ...).
I imagine you could convert dates to floats or integers, build some sort of slider using these, and then convert back to dates for the display labels on the slider. However, this seems clunky. Does anyone have a smoother solution?
A general purpose range slider widget called SelectionRangeSlider has been added to ipywidgets in May 2017. This slider can be used for all sorts of data, including dates.
The code sample below creates a range slider allowing the user to select a date range between April 24 and May 24.
import ipywidgets as widgets
import pandas as pd
from datetime import datetime
start_date = datetime(2018, 4, 24)
end_date = datetime(2018, 5, 24)
dates = pd.date_range(start_date, end_date, freq='D')
options = [(date.strftime(' %d %b %Y '), date) for date in dates]
index = (0, len(options)-1)
selection_range_slider = widgets.SelectionRangeSlider(
options=options,
index=index,
description='Dates',
orientation='horizontal',
layout={'width': '500px'}
)
selection_range_slider
This slider can easily be combined with the interact functionality from ipywidgets. In that case the widget returns a tuple consisting of the selected upper and lower range.
def print_date_range(date_range):
print(date_range)
widgets.interact(
print_date_range,
date_range=selection_range_slider
);
The two outputs for the previously selected ranges:
(Timestamp('2018-04-24 00:00:00', freq='D'), Timestamp('2018-05-24 00:00:00', freq='D'))
(Timestamp('2018-04-30 00:00:00', freq='D'), Timestamp('2018-05-18 00:00:00', freq='D'))
I had the same issue recently. I had to write my own class to do a daterange picker. Here is my code:
import pandas as pd
import ipywidgets as widgets
from IPython.display import display
class DateRangePicker(object):
def __init__(self,start,end,freq='D',fmt='%Y-%m-%d'):
"""
Parameters
----------
start : string or datetime-like
Left bound of the period
end : string or datetime-like
Left bound of the period
freq : string or pandas.DateOffset, default='D'
Frequency strings can have multiples, e.g. '5H'
fmt : string, defauly = '%Y-%m-%d'
Format to use to display the selected period
"""
self.date_range=pd.date_range(start=start,end=end,freq=freq)
options = [(item.strftime(fmt),item) for item in self.date_range]
self.slider_start = widgets.SelectionSlider(
description='start',
options=options,
continuous_update=False
)
self.slider_end = widgets.SelectionSlider(
description='end',
options=options,
continuous_update=False,
value=options[-1][1]
)
self.slider_start.on_trait_change(self.slider_start_changed, 'value')
self.slider_end.on_trait_change(self.slider_end_changed, 'value')
self.widget = widgets.Box(children=[self.slider_start,self.slider_end])
def slider_start_changed(self,key,value):
self.slider_end.value=max(self.slider_start.value,self.slider_end.value)
self._observe(start=self.slider_start.value,end=self.slider_end.value)
def slider_end_changed(self,key,value):
self.slider_start.value=min(self.slider_start.value,self.slider_end.value)
self._observe(start=self.slider_start.value,end=self.slider_end.value)
def display(self):
display(self.slider_start,self.slider_end)
def _observe(self,**kwargs):
if hasattr(self,'observe'):
self.observe(**kwargs)
def fct(start,end):
print start,end
Using it is relatively straightforward:
w=DateRangePicker(start='2016-08-02',end="2016-09-02",freq='D',fmt='%Y-%m-%d')
w.observe=fct
w.display()
Enjoy ;-)
