Loading Pandas Dataframes into Glue-Jupyter.

[Doug Redden]

Hi,

This is similar to a recent question about loading Pandas Dataframes but instead of Glue .... it is for Glue-Jupyter.

I have been loading data from csv files and this works well.   I would now like to load pandas dataframes into Glue-Jupyter.  Please could you give me some advice.  See some of my attempts below

from glue_jupyter.data import require_data

import glue_jupyter as gj

import numpy as np

require_data('Planes/boston_planes_6h.csv')

app = gj.jglue()

planesFromCSV = app.load_data('boston_planes_6h.csv')

df = planes.to_dataframe()

df = df.dropna(how='all', axis=1)  # remove columns with Nans  

df = df.select_dtypes([np.number])  # remove non-numeric columns  

df

# This works well

scatter_viewer = app.scatter2d(x=df.columns[3], y=df.columns[4], data=planesFromCSV)  

# Loading in a dataframe doesn't work

app2 = gj.jglue()

planes_From_df = app2.load_data(df) # KeyError: "Don't know how to open file: 

# planes_From_df = app2.add_data(df)   # KeyError: "Don't know how to open file: 

scatter_viewer = app2.scatter2d(x=df.columns[3], y=df.columns[4], data=planes_From_df, show=False)

Thanks,

Doug

[phed...@gmail.com]

I was only able to load pandas by first saving them as csv files

--
You received this message because you are subscribed to the Google Groups "Glue users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to glue-viz+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/glue-viz/7f7d10e9-ff0e-4551-9332-c100137a94bd%40googlegroups.com.

[Doug Redden]

I've worked it out!  Its quite simple.   See below.

# ------------

from glue_jupyter.data import require_data

import glue_jupyter as gj

import numpy as np

require_data('Planes/boston_planes_6h.csv')

app = gj.jglue()

planesFromCSV = app.load_data('boston_planes_6h.csv')

df = planesFromCSV.to_dataframe()

df = df.dropna(how='all', axis=1)  # remove columns with Nans  

df = df.select_dtypes([np.number])  # remove non-numeric columns  

df

scatter_viewer = app.scatter2d(x=df.columns[3], y=df.columns[4], data=planesFromCSV)  

# Loading in a dataframe doesn't work

app2 = gj.jglue(data=df)

scatter_viewer = app2.scatter2d(x=df.columns[3], y=df.columns[4])

To unsubscribe from this group and stop receiving emails from it, send an email to glue...@googlegroups.com.

[Thomas Robitaille]

Hi Doug,

An alternative solution is to use app.add_data's ability to understand dataframes:

import pandas
df = pandas.read_csv('boston_planes_6h.csv')

planes = app.add_data(planes=df)[0]

After this you can use 'planes' as a regular glue dataset, e.g:

scatter_viewer = app.scatter2d(x='x', y='y', data=planes)

Let me know if this doesn't work!

Cheers,

Tom

To unsubscribe from this group and stop receiving emails from it, send an email to glue-viz+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/glue-viz/d4a98cd1-6bc0-42c0-ba94-de3890a9d1a9%40googlegroups.com.