Change default options in pandas

问题:

I'm wondering if there's any way to change the default display options for pandas. I'd like to change the display formatting as well as the display width each time I run python, eg:

pandas.options.display.width = 150

I see the defaults are hard-coded in pandas.core.config_init. Is there some way in pandas to do this properly? Or if not, is there some way to set up ipython at least to change the config each time I import pandas? Only thing I can think of is making my own mypandas library that wraps pandas with some extra commands issued each time it's loaded. Any better ideas?

回答1:

Have a look at the docs:

Using startup scripts for the python/ipython environment to import pandas and set options makes working with pandas more efficient. To do this, create a .py or .ipy script in the startup directory of the desired profile. An example where the startup folder is in a default ipython profile can be found at:

$IPYTHONDIR/profile_default/startup

More information can be found in the ipython documentation. An example startup script for pandas is displayed below:

import pandas as pd
pd.set_option('display.max_rows', 999)
pd.set_option('precision', 5)

(or use the new form pd.options.display.max_rows = 999).

You also asked:

-- is there any way to only run the pandas code when I import pandas from within ipython? it takes quite a while to import pandas, so I'd rather not do it every time I launch a new ipython instance

As a workaround, you could import pandas in the background. See Import python modules in the background in REPL.

回答2:

As described here, there are iPython config files:

# Most of your config files and extensions will probably start
# with this import

import IPython.ipapi
ip = IPython.ipapi.get()

# You probably want to uncomment this if you did %upgrade -nolegacy
# import ipy_defaults

import os
import pandas


def main():

    #ip.dbg.debugmode = True
    ip.dbg.debug_stack()

    # uncomment if you want to get ipython -p sh behaviour
    # without having to use command line switches
    import ipy_profile_sh
    import jobctrl

    # Configure your favourite editor?
    # Good idea e.g. for %edit os.path.isfile

    #import ipy_editors

    # Choose one of these:

    #ipy_editors.scite()
    #ipy_editors.scite('c:/opt/scite/scite.exe')
    #ipy_editors.komodo()
    #ipy_editors.idle()
    # ... or many others, try 'ipy_editors??' after import to see them

    # Or roll your own:
    #ipy_editors.install_editor("c:/opt/jed +$line $file")


    o = ip.options
    # An example on how to set options
    #o.autocall = 1
    o.system_verbose = 0

    #import_all("os sys")
    #execf('~/_ipython/ns.py')


    # -- prompt
    # A different, more compact set of prompts from the default ones, that
    # always show your current location in the filesystem:

    #o.prompt_in1 = r'\C_LightBlue[\C_LightCyan\Y2\C_LightBlue]\C_Normal\n\C_Green|\#>'
    #o.prompt_in2 = r'.\D: '
    #o.prompt_out = r'[\#] '

    # Try one of these color settings if you can't read the text easily
    # autoexec is a list of IPython commands to execute on startup
    #o.autoexec.append('%colors LightBG')
    #o.autoexec.append('%colors NoColor')
    o.autoexec.append('%colors Linux')

    pandas.options.display.width = 150


# some config helper functions you can use
def import_all(modules):
    """ Usage: import_all("os sys") """
    for m in modules.split():
        ip.ex("from %s import *" % m)

def execf(fname):
    """ Execute a file in user namespace """
    ip.ex('execfile("%s")' % os.path.expanduser(fname))

main()

Probably better to make separate Python profiles. (The code is untested).

回答3:

I was able to fix this by actually going into the pandas folder (found using pandas.__file__). Inside the pandas folder there is a core folder with the config_init.py file. The lines

cf.register_option('large_repr', 'truncate', pc_large_repr_doc,
                       validator=is_one_of_factory(['truncate', 'info']))

set the default options. So you can change the second argument to 'info'

cf.register_option('large_repr', 'info', pc_large_repr_doc,
                       validator=is_one_of_factory(['truncate', 'info']))

and then by default pandas will print summary table if the data frame exceeds max_rows or max_columns, which you can also change the default for in this file. I'm not sure if this is safe behavior, but it worked for me.