CODEWITHSUNDEEP
pythonpandasdataframelambda
user3631926
user3631926

Reputation: 181

pandas add column to dataframe having the value from another row based on condition

I have a dataframe with columns named 'id', 'x', 'y', and 'time'

id time x y
1 0 14 12
1 1 32 23
1 2 52 14
2 2 12 34
3 0 62 17
3 1 82 35
3 2 22 25

I want to add two columns to the dataframe so that they have the value of x and y from another row having the same id and a time + 2

the result should like this:

id time x y x2 y2
1 0 14 12 52 14
1 1 32 23
1 2 52 14
2 2 12 34
3 0 62 17 22 25
3 1 82 35
3 2 22 25

please note that the dataframe is not sorted by id

I have tried the following for x2 but it is not working as intended:

t=2
data['x2'] = data.apply(lambda x: x['x'] if (data[(data['id']==x['id']) & ((data['time']+t) == x['time'])].size > 0) else '', axis=1)

The following works but I need to use a shortcut way and the one with the best performance because my data is huge

t=2
for index, row in data.iterrows():    
    rowT = data[(data['id']==row['id']) & (data['time'] == (row['time'] + t))]
    if rowT.size > 0:
      data.loc[index,'x2'] = rowT['x'].values[0]

Upvotes: 3

Views: 188

Answers (2)

ListenSoftware Louise Ai Agent
ListenSoftware Louise Ai Agent

Reputation: 4233

look up time +2 within each id

id=[1,1,1,2,3,3,3]
time=[0,1,2,2,0,1,2]
x=[14,32,52,12,62,82,22]
y=[12,23,14,34,17,35,25]

df=pd.DataFrame({'id':id,'time':time,'x':x,'y':y})
df.reset_index()
df['x2']=0
df['y2']=0

for key,item in df.iterrows():

   lookup=(item['time']+2) 
   filter=(df['time']==lookup) & (df['id']==item['id'])
   results=df[filter]
   if len(results)>0:
       row=results.iloc[0]
       x2=row.x
       y2=row.y
       df.loc[key,['x2','y2']]=(x2,y2)


print(df) 

output:
   id  time   x   y  x2  y2
0   1     0  14  12  52  14
1   1     1  32  23   0   0
2   1     2  52  14   0   0
3   2     2  12  34   0   0
4   3     0  62  17  22  25
5   3     1  82  35   0   0
6   3     2  22  25   0   0

#no looping
df2=df.copy()
df2['time'] = df2.apply(lambda x: x['time']+2, axis=1)
results=df2[['id','time','x','y']].merge(df[['id','time','x','y']]
,on=['id','time'],how="left",suffixes=('', '2')).fillna(0)
print(results)

Upvotes: 0

Shubham Sharma
Shubham Sharma

Reputation: 71689

You can create a new dataframe by repopulating the values in time column with the values at t-2 seconds, then left merge this new dataframe with the original dataframe on the columns id, time to get the result:

df_r = df.assign(time=df['time'].sub(2))
df.merge(df_r, on=['id', 'time'], how='left', suffixes=['', '2'])

   id  time   x   y    x2    y2
0   1     0  14  12  52.0  14.0
1   1     1  32  23   NaN   NaN
2   1     2  52  14   NaN   NaN
3   2     2  12  34   NaN   NaN
4   3     0  62  17  22.0  25.0
5   3     1  82  35   NaN   NaN
6   3     2  22  25   NaN   NaN

Upvotes: 2

Related Questions