In [ ]:

    

import pandas as pd
pd.__version__


    

In [ ]:

    

fname = '/Users/maye/data/planet4/2014-02-02_planet_four_classifications.csv'


    

In [ ]:

    

df = pd.read_csv(fname, na_values=['null'], nrows=1e6)

# data = [chunk for chunk in reader]
# df = pd.concat(data, ignore_index=True)


    

In [ ]:

    

df.columns


    

In [ ]:

    

df.marking.value_counts()


    

In [ ]:

    

blotches = df[df.marking == 'blotch']


    

In [ ]:

    

fans = df[df.marking == 'fan']


    

In [ ]:

    

s = blotches.iloc[0]


    

In [ ]:

    

s


    

In [ ]:

    

df.marking.unique()


    

In [ ]:

    

def no_of_nulls(row):
    if row.marking in ['interesting', 'none']:
        return True
    if row[row.isnull()].shape[0] !=2:
        return False
    else:
        return True


    

In [ ]:

    

df['okay']=True


    

In [ ]:

    

df['okay'] = df.apply(no_of_nulls, axis=1)


    

In [ ]:

    

df.okay.value_counts()


    

In [ ]:

    

df[df.okay].shape


    

In [ ]:

    

from P4_sandbox import get_data


    

In [ ]:

    

im = get_data.get_image_from_record(fans.loc[357425 ])


    

In [ ]:

    

imshow(im)


    

In [ ]:

    

fans.spread.hist()


    

In [ ]:

    

df.acquisition_date = pd.to_datetime(df.acquisition_date)


    

In [ ]:

    

df.dtypes


    

In [ ]:

    

df.to_hdf?


    

In [ ]:

    

df.to_hdf('/Users/maye/data/planet4/2013-10-27_planet_four_classifications.h5',
          'df')


    

In [ ]:

    

# data_columns=['image_id','image_name','user_name','marking'])