[QST] error 'ndarray' object has no attribute 'to_pandas'

[hakim3189]

Hi All,

recently I have problem using widget in cuxfilter, my code always produce "AttributeError: 'ndarray' object has no attribute 'to_pandas'"
when using Multiselect & Dropdown but working perfectly when using range_slider. Need guidance why this problem happen and how to solve it.

---

AttributeError Traceback (most recent call last)
Cell In[131], line 1
----> 1 d1 = cux_df.dashboard([chart1, chart2],sidebar=[chart3,chart4],layout=cuxfilter.layouts.feature_and_base, theme=cuxfilter.themes.rapids_dark)
2 #[chart1, chart3, chart4], sidebar=[chart2], layout=cuxfilter.layouts.feature_and_double_base, title='Auto Accident Dataset'

File ~/miniconda3/envs/rapids-23.10/lib/python3.10/site-packages/cuxfilter/dataframe.py:236, in DataFrame.dashboard(self, charts, sidebar, layout, theme, title, data_size_widget, warnings, layout_array)
233 if notebook_assets.pn.config.js_files == {}:
234 notebook_assets.load_notebook_assets()
--> 236 return DashBoard(
237 charts=charts,
238 sidebar=sidebar,
239 dataframe=self,
240 layout=layout,
241 theme=theme,
242 title=title,
243 data_size_widget=data_size_widget,
244 show_warnings=warnings,
245 layout_array=layout_array,
246 )

File ~/miniconda3/envs/rapids-23.10/lib/python3.10/site-packages/cuxfilter/dashboard.py:186, in DashBoard.init(self, charts, sidebar, dataframe, layout, theme, title, data_size_widget, show_warnings, layout_array)
184 for chart in sidebar:
185 if chart.is_widget:
--> 186 chart.initiate_chart(self)
187 chart._initialized = True
188 self._sidebar[chart.name] = chart

File ~/miniconda3/envs/rapids-23.10/lib/python3.10/site-packages/cuxfilter/charts/panel_widgets/plots.py:334, in MultiChoice.initiate_chart(self, dashboard_cls)
328 self.min_value, self.max_value = get_min_max(
329 dashboard_cls._cuxfilter_df.data, self.x
330 )
331 self.source = dashboard_cls._cuxfilter_df.data[self.x].reset_index(
332 drop=True
333 )
--> 334 self.calc_list_of_values(dashboard_cls._cuxfilter_df.data)
335 self.generate_widget()
336 self.add_events(dashboard_cls)

File ~/miniconda3/envs/rapids-23.10/lib/python3.10/site-packages/cuxfilter/charts/panel_widgets/plots.py:347, in MultiChoice.calc_list_of_values(self, data)
344 if isinstance(data, dask_cudf.core.DataFrame):
345 self.list_of_values = self.list_of_values.compute()
--> 347 self.list_of_values = self.list_of_values.to_pandas().tolist()
349 if len(self.list_of_values) > 500:
350 print(
351 """It is not recommended to use a column with
352 so many different values for MultiChoice menu"""
353 )

AttributeError: 'ndarray' object has no attribute 'to_pandas'

chart3 = cuxfilter.charts.range_slider('RSRP_Status') --> working
chart3 = cuxfilter.charts.drop_down('RSRP_Status') --> not working

Herewith the dtypes
Tile Id int64
Date datetime64[ns]
Serving Cell Average RSRP (All) (dBm) float64
Serving Cell Average RSRQ (All) (dB) float64
Total MR Count int64
Longitude float64
Latitude float64
x float64
y float64
RSRP_Status int64
dtype: object

[AjayThorve]

Hi @hakim3189, can you please post the versions of cuxfilter, cudf and numpy you are using?

[hakim3189]

herewith the version
cuxfilter version: 23.10.00
cudf version: 23.10.02
numpy version: 1.24.4

[AjayThorve]

Hey @hakim3189, I was not able to reproduce the error on the latest stable release. Can you please give it a try on the version 23.12.* for cudf and cuxfilter?

[hakim3189]

ok i will try to update cudf and cuxfilter first

[hakim3189]

hey @AjayThorve , recently i already update my version.
herewith my latest version
cuxfilter version: 23.12.00
cudf version: 23.12.01
numpy version: 1.23.4

but somehow it still produce this error

---

AttributeError Traceback (most recent call last)
Cell In[88], line 1
----> 1 d1 = cux_df.dashboard([chart1, chart2],sidebar=[chart3,chart4],layout=cuxfilter.layouts.feature_and_base, theme=cuxfilter.themes.rapids_dark)
2 #[chart1, chart3, chart4], sidebar=[chart2], layout=cuxfilter.layouts.feature_and_double_base, title='Auto Accident Dataset'

File ~/miniconda3/envs/rapids-23.12/lib/python3.10/site-packages/cuxfilter/dataframe.py:236, in DataFrame.dashboard(self, charts, sidebar, layout, theme, title, data_size_widget, warnings, layout_array)
233 if notebook_assets.pn.config.js_files == {}:
234 notebook_assets.load_notebook_assets()
--> 236 return DashBoard(
237 charts=charts,
238 sidebar=sidebar,
239 dataframe=self,
240 layout=layout,
241 theme=theme,
242 title=title,
243 data_size_widget=data_size_widget,
244 show_warnings=warnings,
245 layout_array=layout_array,
246 )

File ~/miniconda3/envs/rapids-23.12/lib/python3.10/site-packages/cuxfilter/dashboard.py:186, in DashBoard.init(self, charts, sidebar, dataframe, layout, theme, title, data_size_widget, show_warnings, layout_array)
184 for chart in sidebar:
185 if chart.is_widget:
--> 186 chart.initiate_chart(self)
187 chart._initialized = True
188 self._sidebar[chart.name] = chart

File ~/miniconda3/envs/rapids-23.12/lib/python3.10/site-packages/cuxfilter/charts/panel_widgets/plots.py:334, in MultiChoice.initiate_chart(self, dashboard_cls)
328 self.min_value, self.max_value = get_min_max(
329 dashboard_cls._cuxfilter_df.data, self.x
330 )
331 self.source = dashboard_cls._cuxfilter_df.data[self.x].reset_index(
332 drop=True
333 )
--> 334 self.calc_list_of_values(dashboard_cls._cuxfilter_df.data)
335 self.generate_widget()
336 self.add_events(dashboard_cls)

File ~/miniconda3/envs/rapids-23.12/lib/python3.10/site-packages/cuxfilter/charts/panel_widgets/plots.py:347, in MultiChoice.calc_list_of_values(self, data)
344 if isinstance(data, dask_cudf.core.DataFrame):
345 self.list_of_values = self.list_of_values.compute()
--> 347 self.list_of_values = self.list_of_values.to_pandas().tolist()
349 if len(self.list_of_values) > 500:
350 print(
351 """It is not recommended to use a column with
352 so many different values for MultiChoice menu"""
353 )

AttributeError: 'ndarray' object has no attribute 'to_pandas'

[AjayThorve]

Hey @hakim3189 this is a bit weird, can you get the following working in your enviroment?

import cudf
df = cudf.DataFrame({"a":[1,23,3]}, dtype={"a":"int64"})
print(df['a'].unique().to_pandas().tolist())

the error that's raised refers to the above usage of cudf, which on my 23.12 enviroment, works fine. I would need more info into your data. Also I am assuming you are using this with cuDF and not pandas, is that correct?

[hakim3189]

hey @AjayThorve

it produce this error in my environment

---

AttributeError Traceback (most recent call last)
Cell In[80], line 3
1 import cudf
2 df = cudf.DataFrame({"a":[1,23,3]}, dtype={"a":"int64"})
----> 3 print(df['a'].unique().to_pandas().tolist())

AttributeError: 'ndarray' object has no attribute 'to_pandas'

data is originally in pandas but then I convert to cu_df
cu_df = cudf.from_pandas(df)
cux_df = cuxfilter.DataFrame.from_dataframe(cu_df)

[hakim3189]

Hey @AjayThorve

Finally I found why this happen

at first herewith my Script :
import cuxfilter
%load_ext cudf.pandas
import pandas as pd
from cuxfilter import DataFrame, themes, layouts, charts
from cuxfilter.charts import bokeh
from cuxfilter.charts import datashader
from pathlib import Path
import cudf
import cuml
import datashader as ds
import numpy as np
import geopandas as gpd
import fiona
import cuspatial
from shapely.geometry import Point, Polygon, MultiPolygon
import matplotlib.pyplot as plt

but after I rerun but without this line it work perfectly
%load_ext cudf.pandas

Thanks for your help

[AjayThorve]

Thats great @hakim3189. Yeah I think importing cudf after loading cudf.pandas might have cuased the issue. I'll make sure we make it clear on the usage with cudf.pandas in the documentation soon. Feel free to open up an issue if you face any in future.