In [1]:

    

import json
import pandas as pd
import numpy as np


    

In [16]:

    

hts = pd.read_csv("data/top_hashtags.csv")
# Should have made a better query
hts.head()


    

    
Out[16]:






  

    

      

      
n
      
deg
    
  
  

    

      
0
      
          {"tagid":"h3","hashtag":"paris"}
      
 719571
    
    

      
1
      
   {"tagid":"h2","hashtag":"prayforparis"}
      
 368931
    
    

      
2
      
   {"tagid":"h5","hashtag":"parisattacks"}
      
 206159
    
    

      
3
      
 {"tagid":"h19","hashtag":"prayers4paris"}
      
  52805
    
    

      
4
      
       {"tagid":"h8","hashtag":"bataclan"}
      
  51524
    
  

In [17]:

    

min_deg = hts["deg"].min()
min_deg


    

    
Out[17]:





1798

In [18]:

    

max_deg = hts["deg"].max()
max_deg


    

    
Out[18]:





719571

In [19]:

    

converted = [{"text": "#" + json.loads(row[1]["n"])["hashtag"],
              "size": np.interp(row[1]["deg"], [min_deg, max_deg], [5, 100])} for row in hts.iterrows()]


    

In [20]:

    

f = open('data/hashtag_sizes.js', 'w')
json.dump(converted, f)
f.close()


    

In [ ]: