notebook.community

Edit and run



In [3]:

    
csvfile = "GENdata.csv"



In [4]:

    
csvfile









    Out[4]:





'GENdata.csv'



In [27]:

    
gendata = open(csvfile,'r')



In [28]:

    
gendata









    Out[28]:





<_io.TextIOWrapper name='GENdata.csv' mode='r' encoding='UTF-8'>



In [29]:

    
memdata = gendata.read()



In [30]:

    
memdata









    Out[30]:





'PatientID,GeneticMarker,TestGroup\n1234,sep13,A\n2345,oct14,B\n98754,nov12,A\n323424,mar02,B\n'



In [31]:

    
import pandas as pd



In [34]:

    
csvdf = pd.read_csv(csvfile)

csvdf



In [35]:

    
print csvdf









    



  File "<ipython-input-35-e486b52a4867>", line 1
    print csvdf
              ^
SyntaxError: Missing parentheses in call to 'print'



In [36]:

    
print (csvdf)









    



   PatientID GeneticMarker TestGroup
0       1234         sep13         A
1       2345         oct14         B
2      98754         nov12         A
3     323424         mar02         B



In [37]:

    
csvdf.hist









    Out[37]:





<bound method hist_frame of    PatientID GeneticMarker TestGroup
0       1234         sep13         A
1       2345         oct14         B
2      98754         nov12         A
3     323424         mar02         B>



In [40]:

    
csvdf.TestGroup.astype('category')









    Out[40]:





0    A
1    B
2    A
3    B
Name: TestGroup, dtype: category
Categories (2, object): [A, B]



In [43]:

    
csvhist = csvdf.hist(by=csvdf.TestGroup)



In [44]:

    
csvhist









    Out[44]:





array([<matplotlib.axes._subplots.AxesSubplot object at 0x1138b09b0>,
       <matplotlib.axes._subplots.AxesSubplot object at 0x114fd4cf8>], dtype=object)



In [45]:

    
csvdf.describe









    Out[45]:





<bound method NDFrame.describe of    PatientID GeneticMarker TestGroup
0       1234         sep13         A
1       2345         oct14         B
2      98754         nov12         A
3     323424         mar02         B>



In [ ]: