I have a pandas DataFrame of membership records that contains some entry errors (see table below for a few examples). Some members were incorrectly identified as "Joined" when they were in fact "Renewal" and/or listed as "Joined" multiple times. I want to correct these errors by turning "Joined" into "Renewal" and vice-verse as appropirate based on the column year.
MemTransType2009 | MemTransType2010 | MemTransType2011 | MemTransType2012 | MemTransType2013 | MemTransType2014 | MemTransType2015 |
---|---|---|---|---|---|---|
NaN | NaN | Joined | Renewal | Renewal | Joined | Renewal |
NaN | Joined | Renewal | NaN | Joined | Renewal | Renewal |
Joined | Renewal | Renewal | Renewal | Renewal | Renewal | Renewal |
NaN | NaN | Renewal | Joined | Renewal | Renewal | NaN |
Using np.where and a loop that updates the row I can make corrections column by column. For example:
Years = ['MemTransType2009', 'MemTransType2010', 'MemTransType2011', 'MemTransType2012', 'MemTransType2013', 'MemTransType2014', 'MemTransType2015']
for col in df[Years[1:]]:
df[col] = np.where(((df[Years[0]] == 'Joined') & (df[col] == 'Joined')), 'Renewal', df[col])
for col in TEST[Years[2:]]:
df[col] = np.where(((df[Years[1]] == 'Joined') & (df[col] == 'Joined')), 'Renewal', df[col])
etc...
will get rid of any duplicating "Joined", but I'm not skilled enough to do it for all columns at once. Right now I'm manually updating for each successive column.
I've tried several variations of a more complex loop but I get no response, an error, or accidently overwrite all the data.
For example,
Years = ['MemTransType2009', 'MemTransType2010', 'MemTransType2011', 'MemTransType2012', 'MemTransType2013', 'MemTransType2014', 'MemTransType2015']
for x in range(len(Years)):
for col in df[Years[x+1]]:
df[col] = np.where(((df[Years[x]] == 'Joined') & (df[col] == 'Joined')), 'Renewal', df[col])
Because I have to move forward in time, I believe I need to take the columns sequentially. Is there a way to update the records en masse where values associated with the first column are checked initially and then move on to the next column? Or is there alternative approach?
Thank you for any suggestions / examples.
source https://stackoverflow.com/questions/72764336/conditionally-update-values-in-a-pandas-dataframe-while-moving-across-successive
Comments
Post a Comment