Table of Contents

In [ ]:

    

from planet4 import io


    

In [ ]:

    

db = io.DBManager()


    

In [ ]:

    

db.dbname


    

In [ ]:

    

image_ids = db.image_ids


    

In [ ]:

    

len(image_ids)


    

In [ ]:

    

chosen = np.random.choice(image_ids, 10000)


    

In [ ]:

    

def get_fan_and_blotch_nunique_cids(data):
    f1 = data.marking == 'fan'
    f2 = data.marking == 'blotch'
    return data[f1 | f2].classification_id.nunique()


    

In [ ]:

    

from ipyparallel import Client
c = Client()


    

In [ ]:

    

lbview = c.load_balanced_view()


    

In [ ]:

    

dview = c[:]


    

In [ ]:

    

dview.push({'db': db,
            'get_fan_and_blotch_nunique_cids':get_fan_and_blotch_nunique_cids})


    

In [ ]:

    

def get_ratio(image_id):
    data = db.get_image_id_markings(image_id)
    n_classifications = data.classification_id.nunique()
    n_class_fb = get_fan_and_blotch_nunique_cids(data)
    ratio = (n_class_fb / n_classifications)
    d = {'image_id':image_id, 'ratio':ratio}
    return d


    

In [ ]:

    

results = lbview.map_async(get_ratio, chosen)


    

In [ ]:

    

from nbtools import display_multi_progress


    

In [ ]:

    

display_multi_progress(results, chosen)


    

In [ ]:

    

df = pd.DataFrame(results.result())


    

In [ ]:

    

%matplotlib inline


    

In [ ]:

    

df.ratio.hist(bins=30)


    

In [ ]:

    

import seaborn as sns


    

In [ ]:

    

sns.distplot(df.ratio)


    

In [ ]:

    

t0 = pd.Timestamp('2016-07-01 12:00:00')


    

In [ ]:

    

delta = pd.Timedelta('114h')


    

In [ ]:

    

t0 + delta


    

In [ ]: