CODEWITHSUNDEEP
pythonpandasdata-munging
Viraj Desai
Viraj Desai

Reputation: 75

Filter rows in a dataframe using multiple conditions

Consider a dataframe having 4 columns-

So I am looking to identify rows which are similar on Currency Pair , Notional Quantity, but opposite trade types - Buy and Sell

import pandas as pd


trade_id=[1,2,3,4,5,6] #dtype = int64
ccy_pairs=['AUD','AUD','GBP','EUR','NZD','NZD']#dtype = str
notional=[1,1,1.5,2,6,7]#dtype = int64
trade_type=['buy','sell','buy','sell','buy','buy']#dtype = str
value_date=['01012018', '03012019', '05062018','03062018','07082018','09082020']#dtype = datetime

df=pd.DataFrame() #dataframe comprising of many other columns
df['trade_id']=trade_id
df['ccy_pairs']=ccy_pairs
df['notional']=notional
df['trade_type']=trade_type
df['value_date']=value_date



#Output expected - Looking to highlight the offsetting legs of the trade ( i.e. trades having same notional and ccy pair,
#but different trade types )

Trade Id|CCY Pair|Notional|Trade_type|value_date
1 aud 1 Buy 01012018
3 gbp 1.5 Buy 05062018
4 eur 2 Sell 07062018
5 nzd 6 Buy 07082018
6 nzd 7 Buy 09092020

This means that 2 rows which matched on CCY an Notional but had opposing legs (Buy and Sell) resulted in one of them (either) getting dropped

Upvotes: 2

Views: 111

Answers (2)

ipramusinto
ipramusinto

Reputation: 2668

Condition of two rows to drop (one of them):

"(duplicated rows in ccy_pairs AND notional) AND (not duplicated in trade_type)"

drop_duplicates won't check opposing legs (buy and sell). You can try this (I assume to always drop the second find (.index[1])):

dups = df.ccy_pairs[df.ccy_pairs.duplicated()] # to get AUD and NZD

for i in dups: # to check opposing legs
    if df.trade_type[df.ccy_pairs == i].nunique() == 2:
        df.drop(df[df.ccy_pairs == i].index[1], inplace=True)

Upvotes: 1

Sociopath
Sociopath

Reputation: 13426

You need:

df.drop_duplicates(subset=['ccy_pairs','notional'], keep='first', inplace=True)

output

    trade_id    ccy_pairs   notional    trade_type  value_date
0   1            AUD        1.0          buy         01012018
2   3            GBP        1.5          buy         05062018
3   4            EUR        2.0          sell        03062018
4   5            NZD        6.0          buy         07082018
5   6            NZD        7.0          buy         09082020

For more detail refer this

Upvotes: 1

Related Questions