CODEWITHSUNDEEP
pythonpandaspandas-groupby
Martien Lubberink
Martien Lubberink

Reputation: 2735

How to identify changed values within group - pandas

Is there a simple way to identity groups in a pandas dataframe where a value of a column for a group has changed (or where the group has multiple values for a specific column)?

So for the two companies A and B below, I would like to know which has moved to a different state over time. (Hint, firm A moved from NC to SC).

-------------------
 Yr | Company| ST |  
-------------------
 1  | A      | NC |
 2  | A      | NC |
 3  | A      | SC |
 4  | A      | SC |
 1  | B      | AK |
 2  | B      | AK |
 3  | B      | AK |
------------------

Ideally I would like pandas to create a column that shows "Y" when a firm has moved:

---------------------------
 Yr | Company| ST |Changed|
---------------------------
 1  | A      | NC |    Y
 2  | A      | NC |    Y
 3  | A      | SC |    Y
 4  | A      | SC |    Y
 1  | B      | AK |    N
 2  | B      | AK |    N
 3  | B      | AK |    N
---------------------------

With numbers it is easy, as the mean number in a group will differ from any of the group values. But with text it is harder ... .

Upvotes: 0

Views: 382

Answers (1)

BENY
BENY

Reputation: 323226

Using transform with nunique

df.groupby('Company').ST.transform('nunique').ne(1).map({True:'Y',False:'N'})
Out[418]: 
0    Y
1    Y
2    Y
3    Y
4    N
5    N
6    N
Name: ST, dtype: object
#df['Change']=df.groupby('Company').ST.transform('nunique').ne(1).map({True:'Y',False:'N'})

Upvotes: 3

Related Questions