Issue with Map Plot

[joeborrello]

When I try to run the Mapplot example, I get the following error when I try to use plot_bokeh.map:

AttributeError: 'function' object has no attribute 'map'

The whole code is as follows:

import pandas as pd
import pandas_bokeh
pandas_bokeh.output_notebook()

df_mapplot = pd.read_csv(r"https://raw.githubusercontent.com/PatrikHlobil/Pandas-Bokeh/master/Documentation/Testdata/populated%20places/populated_places.csv")

df_mapplot["size"] = df_mapplot["pop_max"] / 1000000
df_mapplot.plot_bokeh.map(
    x="longitude",
    y="latitude",
    hovertool_string="""<h2> @{name} </h2> 
    
                        <h3> Population: @{pop_max} </h3>""",
    tile_provider="STAMEN_TERRAIN_RETINA",
    size="size", 
    figsize=(900, 600),
    title="World cities with more than 1.000.000 inhabitants")

[PatrikHlobil]

You have to use version 0.1 for the map plots. You can check the version with

import pandas_bokeh
print(pandas_bokeh.__version__)

You can then upgrade via pip:

pip install --upgrade pandas-bokeh

If it works, please close the issue. Otherwise, please provide me the error message such that I can check the problem.

[joeborrello]

Okay, that worked. The originally installed version was 0.0.2, so the pip upgrade fixed that issue. However, I'm not getting the following error:

ValueError: All values of the -columns have to be restricted to (-90, 90). The value corresponds to the longitude in WGS84 projection.

Which doesn't seem to make sense because the data set itself has values that are both above and below 90 and -90, respectively.

[PatrikHlobil]

Thas is weird, I cannot reproduce your problem. I used the code you pasted above (works both with output_notebook and output_file):

import pandas as pd
import pandas_bokeh
pandas_bokeh.output_file("mapplot.html")

df_mapplot = pd.read_csv(r"https://raw.githubusercontent.com/PatrikHlobil/Pandas-Bokeh/master/Documentation/Testdata/populated%20places/populated_places.csv")

df_mapplot["size"] = df_mapplot["pop_max"] / 1000000
df_mapplot.plot_bokeh.map(
    x="longitude",
    y="latitude",
    hovertool_string="""<h2> @{name} </h2> 
    
                        <h3> Population: @{pop_max} </h3>""",
    tile_provider="STAMEN_TERRAIN_RETINA",
    size="size", 
    figsize=(900, 600),
    title="World cities with more than 1.000.000 inhabitants")

that creates the plot below on my computer:
mapplot.zip

The error message comes from restriction of the latitude (this is actually an error in the ValueError-message since it is not the longitude restriction) of the WGS84 coordinates. The data however does not contain values that exceed this boundaries:

df_mapplot.describe()

	pop_max	latitude	longitude	size
count	5.000000e+02	500.000000	500.000000	500.000000
mean	2.786597e+06	25.000442	37.206103	2.786597
std	3.108616e+06	21.067590	77.569481	3.108616
min	1.002000e+06	-37.818085	-123.123590	1.002000
25%	1.283250e+06	17.247772	-7.031884	1.283250
50%	1.720500e+06	30.661972	46.535048	1.720500
75%	2.920250e+06	38.591966	110.388075	2.920250
max	3.567600e+07	60.177509	174.763027	35.676000

which created the plot. Can you check your DataFrame with the describe() method?

[joeborrello]

Okay, so one set of apt updates, and upgrades later, the example code works properly. But I'm still getting that error with a data set of my own. Here's what happens when I use describe() on that data frame:

_	Latitude	Longitude
count	663784.000000	663784.000000
mean	40.765353	-73.921533
std	0.089083	0.050768
min	40.499440	-74.250642
25%	40.680547	-73.953867
50%	40.786001	-73.924239
75%	40.845376	-73.889728
max	40.911587	-73.701025

[PatrikHlobil]

Could you provide me your dataframe (e.g. as pickle or as parquet)? Then I could try it on my computer and figure out what the problem is. Otherwise, I would propose that you try:
df_mapplot.head(100).plot_bokeh.map(...)

and if that works you can try to increase the number of rows that you use for the plot to figure out when the problem occurs.

[joeborrello]

Okay, so using head(100) worked, so maybe there's just too many rows in the data set? It is fairly large data from NYC. I'm including the dataframe also: https://we.tl/t-FBr853It7n

[PatrikHlobil]

Ok, I found the problem. Your x and y columns contain NaN values. Therefore, checking the min/max values gives NaN and this violates the conditions for latitude and longitude. One should implement a warning that one cannot plot NaN values.

The solution for you would be to filter out the rows with NaN values in Latitude or Longitude:

df_mapplot = df_mapplot[df_mapplot["Latitude"].notnull()&df_mapplot["Longitude"].notnull()]

and then to plot the data. Note however that you shouldn't use output_notebook, but output_file. On my computer the notebook crashed because the output was to big. Also, 600.000 points will probably crash your browser, even when using WebGL which is the default for Pandas Bokeh.

[joeborrello]

Thanks for helping with that - I thought I had gotten rid of all of the NaN values, but I guess not. And thanks for the heads up about notebook versus file output.

[PatrikHlobil]

You’re welcome. Maybe have a look at the Output and Layouts section of the README.md. This could also be interesting for you...

[Rendiere]

Hey @PatrikHlobil, I'm running into a potentially related issue on the map plot example.

Using the exact code as in the github readme, I get the error:

AttributeError: 'FramePlotMethods' object has no attribute '_data'

Version:

print(pandas_bokeh.__version__)
> 0.1.1

[PatrikHlobil]

Hi @Rendiere ,

thanks for the bug report. This happens when you don't catch up with all your tests :(.

Thisis a new bug and is different from the one desribed above. I already fixed it and it will be available with the next release.

If you don't want to wait, just upgrade on the dev version via:

pip install git+https://github.com/PatrikHlobil/Pandas-Bokeh.git

Best,

Patrik

[boustrephon]

Hi @PatrikHlobil : I am on 0.3 (from conda on a Mac) and I have the same problem as @Rendiere (AttributeError: 'FramePlotMethods' object has no attribute '_data'). I am using the same code as above.

Python version 3.7.7 (default, Mar 23 2020, 17:31:31)
[Clang 4.0.1 (tags/RELEASE_401/final)]
Pandas version 0.23.4
Bokeh version 2.0.1
Pandas-bokeh version 0.3

[PatrikHlobil]

Hi @boustrephon,

The conda package is out of date at the moment because there are problems with a dependency relation in used packages. Please install the latest version via pip and report If the issue I solved.

Best Patrik

[PatrikHlobil]

Hi @boustrephon, @joeborrello,

I just fixed the conda build problem and you can now also install the most recent version of pandas-bokeh (0.4.3) via conda via:

conda install -c patrikhlobil pandas-bokeh

Best Patrik