Hvplot.points rendering triangle instead of individual points

Hello all, so I have a simple pandas DataFrame with raw longitude and latitude values, and am simply trying to plot using hvplot’s API. All my code is the following:

import pandas as pd # edit
import geopandas as gpd # edit
import geoviews as gv # edit
df = load_some_dataset # edit / placeholder


Instead of individual points on a random subset of my test dataset however, I get a triangle like the following:

which happens to encompass the points I am interested in but not show their individual location. How can I address this?

EDIT: The issue persists even if I manually build the same using GeoViews:

ds = gv.Dataset(df,kdims=["longitude","latitude"]).to(gv.Points,["longitude","latitude"],[])
tiles = gv.tile_sources.OSM

tiles * ds

or converting the dataframe to GeoPandas and then plotting, which still gives me this strange Triangle

gdf = gpd.GeoDataFrame(df, geometry=gpd.points_from_xy(df["longitude"],df["latitude"]))

EDIT 2: After further testing, this seems to only happen in a Jupyter notebook, but renders correctly when I launch a Panel server. The plot renders as a Bokeh one in Jupyter it just seems strange that a discrepancy exists.

Can you provide some data? I’m unable to reproduce.

Sure thing, here is a small sample of the pandas DataFrame in CSV form (index is NOT included). Also I added a second EDIT describing how this only happens in the Jupyter notebook (which renders in Bokeh) but not the Panel server app which renders correctly.

sample_df.csv (8.1 KB)

I can’t reproduce with that data; also maybe change your alpha to float

import pandas as pd
import hvplot.pandas

df = pd.read_csv("sample_df.csv")
    "longitude", "latitude", geo=True, tiles="OSM", color="blue", alpha=0.1, global_extent=True

@ahuang11 this thread slipped from my attention but I wanted to mention that after restarting my IDE that was running the notebook, it all of a sudden rendered correctly and has continued to do so. Note, I fiddled around at the time with packages and may have corrected a jupyter-bokeh installation as well.