pandas .ix slicing deprecated - how to replace?

Question

Executing my Python code containing pandas df slicing with .ix, I get a warning about its future deprecation. I have been trying to find a replacement. But can for the life of me not get the code to work.

I have followed this answer pandas iloc vs ix vs loc explanation, how are they different? and many more

If you try to remove the 2 lines with .loc slicing, and replace it with the line above that uses .ix, you may see the code work properly.

Please help.

My code that fetches updated excange-rates from the US Federal Reserve Bank.

module = __name__

import pandas as pd
pd.core.common.is_list_like = pd.api.types.is_list_like
from pandas_datareader import data as web
import numpy as np
import datetime

xratelist = ['DEXDNUS', 'DEXUSEU']
xrts = []


def xRateList_pd(xratelist, modus='sim'):
    years = 1.2
    days = int(252 * years)  # ant. arb. dage pr år = 252

    if modus == 'sim':
        start   = datetime.datetime(2016,1,1)  # indstil manuelt
        end     = datetime.datetime(2018,5,18) # indstil manuelt

    if modus == 'trading':
        end     = pd.Timestamp.utcnow()
        start   = end - days * pd.tseries.offsets.BDay()

    # Pick all rates simultaneously
    print('Fetching xratelist from Fred: ', xratelist)
    for xrt in xratelist:
        r = web.DataReader(xrt, 'fred',
                       start = start, end = end)
        # add a symbol column
        xrts.append(r)
    # concatenate all the dfs into one
    df_xrates = pd.concat(xrts, axis='columns')  

    df_xrates['DEXDNUS'] = np.round(df_xrates['DEXDNUS'], decimals=4)
    df_xrates['DEXUSEU'] = np.round(df_xrates['DEXUSEU'], decimals=4)


#    xrateDNUS = float(df_xrates.ix[-1, 'DEXDNUS'])  # Works but deprecated...
    xrateDNUS = float(df_xrates.loc[-1, 'DEXDNUS'])   # Finds last x-rate in pd
    xratedateDNUS = df_xrates.index[-1]  # Finds date for this xrate, index..
    print('Found xrateDNUS DKK/USD    : %1.2f DKK, from: %s' %(xrateDNUS, xratedateDNUS))

#    xrateUSEU = float(df_xrates.ix[-1, 'DEXUSEU'])  # Works but deprecated...
    xrateUSEU = float(df_xrates.loc[-1, 'DEXUSEU'])
    xratedateUSEU = df_xrates.index[-1]
    print('Found xrateUSEU USD/EUR    : %1.2f $, from: %s' %(xrateUSEU, xratedateUSEU))

    return df_xrates, xrateDNUS, xratedateDNUS, xrateUSEU, xratedateUSEU


try:
    ###  Get Xrates  =============================================================

    print('')
    print('                ==== $ ====')
    print('')
    print('Looking up DKK/USD & USD/EUR online (US Federal Reserve Bank)')  # via Federal Res. Bank (using pd DataReader)'


    df_xrates, xrateDNUS, xratedateDNUS, xrateUSEU, xratedateUSEU = xRateList_pd(xratelist, modus='trading') #,years=0.25 ,start=datetime.datetime(2000,1,1), end=pd.Timestamp.utcnow()


except: 
    print('!!! Error in module %s !!!' %(module))
    pass


if __name__ == '__main__':

    print("== m03_get_x_rates == is being run directly")
else:
    print("Importing and processing == %s == from CrystallBall" %(module))

Here's the official documentation on how to replace that: https://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated — Vivek Kumar, Jul 20 '18 at 13:56

score 0 · Accepted Answer · answered Jul 21 '18 at 08:06

Thanks guys, I am well aware of the possibilities of .iloc as I wrote in the question. But I havent been able to crack how to work it in my example.

OK - Finally after 2 days I cracked the nut. I found a new way of slicing a df with .iat and this works! Phew.

So this deprecated line of slicing with .ix

xrateDNUS = float(df_xrates.ix[-1, 'DEXDNUS'])  # Works but deprecated...

is replaced with this more modern way of slicing with .iat

xrateDNUS = float(df_xrates.iat[-1, 0])   # Finds last x-rate in pd

and it works.

Actually I also with the same slicing referencing syntax got .iloc to work

xrateDNUS = float(df_xrates.loc[-1, 0])   # Finds last x-rate in pd

Strange that I cannot get .loc to work with my column labels... I wonder...

xrateDNUS = float(df_xrates.iloc[-1, 'DEXDNUS'])

If it makes sense to anyone why .iloc shouldn't work please comment....

pandas .ix slicing deprecated - how to replace?

1 Answers1