CODEWITHSUNDEEP
pythonpandasdataframe
Msquare
Msquare

Reputation: 835

python combining 26 dataframes of different timestamps and same columns

A database in US receives one day information from a place around the world in .csv format. There are 15 days of data and coming from 26 places. A total of 15x26 = 390 data frames. In addition, 26 places have a common reference data frame. Now, I want to combine 390 + 1 reference dataframes into one. I have given here a sample of my problem: 

plA_d1df =        ### place A day1 dataframe
                       Adata
2019-01-10 07:35:22   10
2019-01-10 08:15:17   20

plB_d1df = 
                       Bdata
2019-01-10 07:38:45   30
2019-01-10 08:18:57   40

ptA_d2df = 
                       Adata
2019-01-21 07:35:42   50
2019-01-21 08:15:17   60

ptB_d2df = 
                       Bdata
2019-01-21 07:39:04   70
2019-01-21 08:19:22   80

reference = 
                          ref
2019-01-10 07:35:00     500
2019-01-10 07:38:00     530
2019-01-10 08:15:00     560
2019-01-10 08:18:00     590
2019-01-21 07:35:00     610
2019-01-21 07:39:00     640
2019-01-21 08:15:00     670
2019-01-21 08:19:00     700

Above data of all places and reference should be combined to the timestamp of place-A as given below: 

combdf = 
  datetime            ref0  Adata     ref1   Bdata  
2019-01-10 07:35:22    500   10      530    30
2019-01-10 08:15:17    560   20      590    40  
2019-01-21 07:35:42    610   50      640    70
2019-01-21 08:15:17    670   60      700    80

I implemented following code after referring the solved answer: 

biglist = [[plA_d1df,plB_d1df],[plA_d2df,plB_d2df]] ## dataframes are in a nested list of list
l = []
s1 = []
### refdf = reference dataframe
for i in range(0,len(biglist),1):
    for j in range(0,len(biglist[i]),1):       
            s1=refdf.reindex(biglist[i][j].index,method='nearest')            
        if j==0:
            l.append(s1.join(biglist[i][j]))
        else:
            l.append(s1.join(biglist3[i][j]).reindex(l[0].index,method='nearest'))
combdf = pd.concat(l,1)

Above code ran successfully. Timestamp of combined dataframe combdf matches with place A, which is what I wanted. But the columns of same place did not merge. Instead, seperate columns were created for each day. So I eneded up having 8 columns, instead 4, mostly filled with nan. My present output is: 

combdf = 
  datetime            ref0  Adata   ref1   Bdata   ref0  Adata   ref1   Bdata  
2019-01-10 07:35:22    500   10     530    30       nan    ..          nan
2019-01-10 08:15:17    560   20     590    40       nan    ..          nan
2019-01-21 07:35:42    nan    ..          nan       610   50     640    70
2019-01-21 08:15:17   nan    ..          nan        670   60     700    80

What corrections I have to make to merge columns into same.

Upvotes: 2

Views: 127

Answers (1)

BENY
BENY

Reputation: 323326

Change your code to 

biglist = [[df1,df2],[df3,df4]] ## dataframes are in a nested list of list
l = []
s1 = []
for i in range(0,len(biglist),1):
    l1=[]
    for j in range(0,len(biglist[i]),1):
            s1=refdf.reindex(biglist[i][j].index,method='nearest')
            if j==0:
                l1.append(s1.join(biglist[i][j]))
            else:
                l1.append(s1.join(biglist[i][j]).reindex(l1[0].index,method='nearest'))
    l.append(pd.concat(l1,axis=1))
combdf = pd.concat(l,0)
combdf
Out[252]: 
                     ref  Adata  ref  Bdata
2019-01-10 07:35:22  500     10  530     30
2019-01-10 08:15:17  560     20  590     40
2019-01-21 07:35:42  610     50  640     70
2019-01-21 08:15:17  670     60  700     80

Upvotes: 2

Related Questions