Mistake in matching basemap through contextily

Question

I am focusing on visualizing some datasets of Switzerland through 'contextily' package in python. (Here is just the sample. In fact, the whole dataset has 16000 records in Switzerland.)

	Unnamed: 0	User_location_code	Twitter_created_time	TRUE
16	226333	CH	1/1/2018	(46.2017559, 6.1466014)
23	4488879	CH	1/1/2018	(47.3744489, 8.5410422)
28	3772151	CH	1/1/2018	(47.1486137, 8.5539378)
35	3226920	CH	1/1/2018	(46.4978958, 9.8392428)
51	4577072	CH	1/1/2018	(46.2131843, 6.0815756)
55	4464950	CH	1/1/2018	(46.4602978, 6.8418655)
65	173633	CH	1/1/2018	(46.2767572, 6.168958)
86	4670064	CH	1/1/2018	(46.2922522, 7.5323195)
89	529343	CH	1/1/2018	(46.2017559, 6.1466014)
90	879080	CH	1/2/2018	(47.2617701, 8.4197059)
100	4195443	CH	1/2/2018	(46.2017559, 6.1466014)
106	1837089	CH	1/2/2018	(46.2017559, 6.1466014)
109	4392712	CH	1/2/2018	(46.4705779, 6.8436858)
110	4191284	CH	1/2/2018	(46.6932583, 6.4611912)
112	3581491	CH	1/2/2018	(47.3744489, 8.5410422)
129	4238370	CH	1/3/2018	(46.5332403, 9.874586)
146	3103734	CH	1/4/2018	(46.4602978, 6.8418655)
147	3103733	CH	1/4/2018	(46.4602978, 6.8418655)
179	4531534	CH	1/6/2018	(47.4634767, 7.8125639)
184	207311	CH	1/6/2018	(47.0505452, 8.3054682)
194	974817	CH	1/6/2018	(46.2017559, 6.1466014)
205	4623486	CH	1/7/2018	(46.4978958, 9.8392428)
214	2782686	CH	1/7/2018	(46.9886931, 6.8906893)
221	4394683	CH	1/8/2018	(46.5218269, 6.6327025)
223	586320	CH	1/8/2018	(47.31874, 7.6698284)
229	787444	CH	1/8/2018	(47.4991723, 8.7291498)
238	1615538	CH	1/9/2018	(46.6356963, 6.5320717)
248	543950	CH	1/9/2018	(47.4250593, 9.3765878)
249	1005935	CH	1/9/2018	(47.3744489, 8.5410422)
252	4217738	CH	1/10/2018	(46.1839156, 6.1224047)
269	411809	CH	1/11/2018	(46.2017559, 6.1466014)
271	1661548	CH	1/11/2018	(47.33490645, 9.163183508014553)
279	814333	CH	1/12/2018	(46.6356963, 6.5320717)
286	652801	CH	1/12/2018	(47.3744489, 8.5410422)
293	591620	CH	1/13/2018	(46.2231863, 6.2242767)
297	4378107	CH	1/13/2018	(46.4978958, 9.8392428)
301	2356500	CH	1/13/2018	(46.5939043, 7.9078016)
308	652799	CH	1/13/2018	(47.3744489, 8.5410422)
315	649442	CH	1/14/2018	(46.6583588, 7.6965086)
319	4380071	CH	1/14/2018	(46.2017559, 6.1466014)
320	190831	CH	1/14/2018	(46.5032766, 6.6867879)
321	4384644	CH	1/14/2018	(46.4602978, 6.8418655)
325	4629311	CH	1/14/2018	(46.647529399999996, 7.080368826694413)
339	411808	CH	1/16/2018	(46.2017559, 6.1466014)
341	2783072	CH	1/16/2018	(46.5218269, 6.6327025)
344	4395955	CH	1/16/2018	(46.6789116, 7.1027113)
348	4394190	CH	1/16/2018	(46.2017559, 6.1466014)
354	440279	CH	1/17/2018	(46.7410614, 7.2193275)
358	4397447	CH	1/17/2018	(46.7899498, 7.2183078)

Because of the geographical coordinate in meter, I decided to use the 'EPSG:3857' as the CRS.

# Load the environment
import numpy
import pandas
import geopandas
import pysal
import seaborn
import contextily
import matplotlib.pyplot as plt

# Read the dataset
CH_points = gpd.read_file('CH_points04.shp')

# Print out the crs of the dataset
print(CH_points.crs)

CH_points_3857 = CH_points.to_crs("epsg:3857")

# Print out the crs of the dataset
print(CH_points_3857.crs)

However, I met a mistake during the basemap time. Specifically, in the following code, I have finished the preliminary visualization about the dataset that I used.

# Generate scatter plot for quick checking
seaborn.jointplot(x = "Longitude", y = "Latitude", data = CH_points, s = 2.0);

The result is here, it works well.

And then, I consider using the function 'contextily.add_basemap()' to make kde-plot looks better, at least in Switzerland.

# Set up figure and axis
f, ax = plt.subplots(1, figsize=(10, 10))
seaborn.kdeplot(
    CH_points_3857["X"],
    CH_points_3857["Y"],
    n_levels=50,
    shade=True,
    alpha=0.55,
    cmap="viridis_r",
)
# Add basemap
contextily.add_basemap(
    ax, crs = CH_points_3857.crs.to_string(), source=contextily.providers.OpenStreetMap.CH
)
# Remove axes
ax.set_axis_off()

The result is here. As you see, it can load the basemap, not bad, right? But the basemap now is in France, not in Switzerland.

How can I align these properly?

Note: For pictures, you could see here. https://github.com/geopandas/contextily/issues/212

For the dataset, you could find it here. https://drive.google.com/file/d/1c85ofyiHKDxjyV5m9LgvWNWL_E9fiifb/view?usp=sharing