NYC Taxi Example Slightly Reworked but inoperable

Can you tell me if this is a coding problem or an issue? I seem to always have my heat map show up near NULL Island. I figured it was a projection problem, so tried to work this per the gv.Points method as well and reproject the data (like some NYC taxi examples). Have I misapplied the code or did I find a new projection issue? Thanks.

import param, parambokeh
from colorcet import cm_n
from holoviews.streams import RangeXY
from bokeh.tile_providers import STAMEN_TONER

#url='https://server.arcgisonline.com/ArcGIS/rest/services/World_Imagery/MapServer/tile/{Z}/{X}/{Y}.jpg'
#url='https://maps.wikimedia.org/osm-intl/{Z}/{X}/{Y}.png'
#url='http://c.tile.openstreetmap.org/{Z}/{X}/{Y}.png'
url=STAMEN_TONER
tiles = gv.WMTS(url,crs=crs.GOOGLE_MERCATOR)
opts = dict(width=1000,height=600,xaxis=None,yaxis=None,bgcolor='black',show_grid=False)

class AirTraffic(hv.streams.Stream):
    alpha      = param.Magnitude(default=0.75, doc="Alpha value for the map opacity")
    colormap   = param.ObjectSelector(default=cm_n["fire"], objects=cm_n.values())

    def make_view(self, x_range, y_range, **kwargs):
        map_tiles = tiles.options(alpha=self.alpha, **opts)
        points = hv.Points(ddf, ['lon', 'lat'])
        taxi_trips = datashade(points, cmap=self.colormap,
                               dynamic=False, x_range=x_range, y_range=y_range, width=1000, height=600)
        return map_tiles * taxi_trips

explorer = AirTraffic(name="Russell Traffic")
parambokeh.Widgets(explorer, callback=explorer.event)
hv.DynamicMap(explorer.make_view, streams=[explorer, RangeXY()])

Thanks.

[philippjfr]

I noticed that the app update cycle takes quite some time (~700k rows) as I never was able to see it update (even after several minutes) with the map tile added into the widget. I was using a 8vCPUs 16GB RAM GCP Ubuntu machine.

I would suggest projecting the points ahead of time, with:

points = gv.Points(ddf, ['lon', 'lat'])
projected_points = gv.project(points)

and then datashade that.

[jbednar]

Hi Tyler! Based on your plot above, datashader does seem like what your data deserves; just about any other way of showing that data will obscure all the detail that you can see there.

Ordinarily, the tiles shouldn’t affect the display speed measurably. But the tiles only support one projection (Web Mercator), and so if you overlay on tiles then your points have to be reprojected (expensively) before datashader can do its aggregations, every time you compute a plot. So for interactive use you should probably project your points all at once, making subsequent datashading on a tile background much quicker. @philippjfr can probably suggest a more convenient way, but what I use for reprojecting is the simple lnglat_to_meters() function in datashader.util.