Python Forum
Pandas referring to result of a previous row - Printable Version

+- Python Forum (https://python-forum.io)
+-- Forum: Python Coding (https://python-forum.io/forum-7.html)
+--- Forum: Data Science (https://python-forum.io/forum-44.html)
+--- Thread: Pandas referring to result of a previous row (/thread-13084.html)

Pandas referring to result of a previous row - slavok400 - Sep-27-2018

Hi people,

Would really appreciate if you help me with something.
You know in Excel when you apply a formula on a cell and then drag it down so every row would be dependent on the result of the cell right above it?

So I need to do it in Python somehow.

I found a solution, but it's not working on my training set. Please advise!
Thanks alot for your help (=

I know with .shift() I can refer to a previous cell if they are not in the same column, .rolling() is providing good, but limited options. Also I cannot use cumulative functions, cause the result will differ depending on other columns' output.

I think my best shot is going for .apply()

df = pd.DataFrame([{'a': 15, 'b': 15, 'Cash': 5}, {'a': 20, 'b': 10, 'Cash': 7}, {'a': 25, 'b': 30, 'Cash': 9}])

def div(x):    
    last_row_id = x.name - 1 
    
    if x.name == 0:
        x['Cash'] = 1       
    else:      
        x['Cash'] = df['Cash'][last_row_id] + 1    
    return x

df.apply(div,axis=1)
What I get from it:
Output:
	Cash	a	b
0	1	15	15
1	2	20	10
2	3	25	30
BUT as soon as I apply this logic to a bigger database, it stops working. Here is my problem:

test = result.copy()
test.reset_index(inplace=True)

def div(x):
    
    last_row_id = x.name - 1
    
    if x.name == 0:
        x['Cash']= 1
    else:
        x['Cash'] = test['Cash'][last_row_id] + 1
    
    return x

test = test.apply(div,axis=1)
test
Cash is supposed to go like 1,2,3,4,5, but it's not. 

Output:
    Cash	  QQQ
0	1	157.611725
1	1	159.143173
2	1	159.421616
3	1	161.022705
4	1	161.649200
5	1	161.659149


RE: Pandas referring to result of a previous row - slavok400 - Sep-27-2018

SOLVED.

I used "for i, data in df.iterrows()" method