1

I'm trying to access filtered versions of a dataframe, using a list with the filter values.

I'm using a while loop that I thought would plug the appropriate list values into a dataframe filter one by one. This code prints the first one fine but then prints 4 empty dataframes afterwards.

I'm sure this is a quick fix but I haven't been able to find it.

boatID = [342, 343, 344, 345, 346]
i = 0 
while i < len(boatID):
    df = df[(df['boat_id']==boatID[i])]
    #run some code, i'm printing DF.head to test it works
    print(df.head())
    i = i + 1

Example dataframe: 

   boat_id  activity speed  heading
0      342         1  3.34   270.00
1      343         1  0.02     0.00
2      344         1  0.01   270.00
3      345         1  8.41   293.36
4      346         1  0.03    90.00
filmor
  • 30,840
  • 6
  • 50
  • 48
hselbie
  • 1,749
  • 9
  • 24
  • 40
  • Thanks for the suggestion, i'm not trying to return a bool value generated by `isin`, I'm trying to filter the DF where `boat_id` == some number. – hselbie Jan 22 '16 at 23:38
  • Update, using `int(boatID[i])` doesn't work either – hselbie Jan 22 '16 at 23:39

1 Answers1

1

I think you overwrite df by df in df = df[(df['boat_id']==boatID[i])]:

Maybe you need change output to new dataframe, e.g. df1:

boatID = [342, 343, 344, 345, 346]
i = 0 
while i < len(boatID):
    df1 = df[(df['boat_id']==boatID[i])]
    #run some code, i'm printing DF.head to test it works
    print(df1.head())
    i = i + 1

#   boat_id  activity  speed  heading
#0      342         1   3.34      270
#   boat_id  activity  speed  heading
#1      343         1   0.02        0
#   boat_id  activity  speed  heading
#2      344         1   0.01      270
#   boat_id  activity  speed  heading
#3      345         1   8.41   293.36
#   boat_id  activity  speed  heading
#4      346         1   0.03       90

If you need filter dataframe df with column boat_id by list boatID use isin:

df1 = df[(df['boat_id'].isin(boatID))]
print df1
#   boat_id  activity  speed  heading
#0      342         1   3.34   270.00
#1      343         1   0.02     0.00
#2      344         1   0.01   270.00
#3      345         1   8.41   293.36
#4      346         1   0.03    90.00

EDIT:

I think you can use dictionary of dataframes:

print df
   boat_id  activity  speed  heading
0      342         1   3.34   270.00
1      343         1   0.02     0.00
2      344         1   0.01   270.00
3      345         1   8.41   293.36
4      346         1   0.03    90.00

boatID = [342, 343, 344, 345, 346]

dfs = ['df' + str(x) for x in boatID]
dicdf = dict()

print dfs
['df342', 'df343', 'df344', 'df345', 'df346']

i = 0 
while i < len(boatID):
    print dfs[i]
    dicdf[dfs[i]] = df[(df['boat_id']==boatID[i])]
    #run some code, i'm printing DF.head to test it works
#    print(df1.head())
    i = i + 1

print dicdf
{'df344':    boat_id  activity  speed  heading
2      344         1   0.01      270, 'df345':    boat_id  activity  speed  heading
3      345         1   8.41   293.36, 'df346':    boat_id  activity  speed  heading
4      346         1   0.03       90, 'df342':    boat_id  activity  speed  heading
0      342         1   3.34      270, 'df343':    boat_id  activity  speed  heading
1      343         1   0.02        0}

print dicdf['df342']
   boat_id  activity  speed  heading
0      342         1   3.34      270
Community
  • 1
  • 1
jezrael
  • 822,522
  • 95
  • 1,334
  • 1,252
  • thanks, this looks like it concatenates all the DF's into 5 dtype objects, in one DF, is that correct? – hselbie Jan 23 '16 at 00:08
  • Now it creates 5 new dataframes `df1` from `df`. In next loop old `df1` is overwritten new `df1` – jezrael Jan 23 '16 at 00:17
  • Is it possible rather than have df1 overwritten each time the loop runs, have something like `df[i]` to create 5 dataframes with a unique name like df342, df343 etc? – hselbie Jan 25 '16 at 17:56