notebook.community

Edit and run 





In [22]:



    


from pprint import pprint
import json

name = 'third-batch';

with open('{}.json'.format(name)) as f:
    data = json.load(f)['logs']
    
with open('images/_airlines.json') as f:
    airlines = json.load(f)

with open('images/_states.json') as f:
    states = json.load(f)



    











In [23]:



    


from collections import OrderedDict 
from copy import deepcopy

rows = []

for p in data.items():
    user = p[0]
    d = OrderedDict({
        'id': user
    })
    
    item = p[1]
    
    if not 'done' in item:
        print('ignoring {}'.format(p[0]))
        continue
    
    d['userAgent'] = item['userAgent']
    d['isMobile'] = item['isMobile']
    
    cond = item['cond']
    
    errorAirline = cond['errorAirline']
    seqAirline = cond['seqAirline']
    errorStates = cond['errorStates']
    seqStates = cond['seqStates']
    focusAirline = cond['focusAirline']
    focusState = cond['focusState']
    
    for k, v in cond.items():
        d[k] = v
        
    howManyIndex = 0
    for k, v in item['approximateVis_airline'].items():
        if k == 'done':
            continue
        t = v['type']
        
        if not 'confidence' in v:
            raise Exception(user, v)
        
        approx = airlines["{}_{}".format(errorAirline, seqAirline)]
        
        prefix = 'approx.airline.'
        
        if (t == 'HowMany'):
            p = prefix + 'howMany.'
            d[p + 'answer'] = int(v['answer'])
            d[p + 'approx'] = approx[focusAirline]
            d[p + 'confidence'] = v['confidence']
            
        if (t == 'HowManyCompare'):
            p = prefix + 'howManyCompare.' + str(howManyIndex) + '.'
            d[p + 'answer'] = int(v['answer'])
            d[p + 'data'] = v['data']
            d[p + 'approx'] = approx[v['data']]
            d[p + 'confidence'] = v['confidence']
            howManyIndex += 1
    
    howManyIndex = 0
    for k, v in item['approximateVis_states'].items():
        if k == 'done':
            continue
        t = v['type']
        
        if not 'confidence' in v:
            raise Exception(user, v)
        
        approx = states["{}_{}".format(errorStates, seqStates)]
        
        prefix = 'approx.states.'
        
        if (t == 'HowMany'):
            p = prefix + 'howMany.'
            d[p + 'answer'] = int(v['answer'])
            d[p + 'approx'] = approx[focusState]
            d[p + 'confidence'] = v['confidence']
            
        if (t == 'HowManyCompare'):
            p = prefix + 'howManyCompare.' + str(howManyIndex) + '.'
            d[p + 'answer'] = int(v['answer'])
            d[p + 'data'] = v['data']
            d[p + 'approx'] = approx[v['data']]
            d[p + 'confidence'] = v['confidence']
            howManyIndex += 1
     
    def jaccard_similarity(x,y):
        intersection_cardinality = len(set.intersection(*[set(x), set(y)]))
        union_cardinality = len(set.union(*[set(x), set(y)]))
        return intersection_cardinality/float(union_cardinality)
    
    # precise
            
    howManyIndex = 0
    for k, v in item['preciseVis_airline'].items():
        if k == 'done':
            continue
        t = v['type']
        
        if not 'confidence' in v:
            raise Exception(user, v)
        
        precise = airlines["{}_{}".format(1, 1)]
        approx = airlines["{}_{}".format(errorAirline, seqAirline)]
        
        prefix = 'precise.airline.'
        
        if (t == 'HowMany'):
            p = prefix + 'howMany.'
            d[p + 'answer'] = int(v['answer'])
            d[p + 'approx'] = approx[focusAirline]
            d[p + 'precise'] = precise[focusAirline]
            d[p + 'confidence'] = v['confidence']
            
        if (t == 'HowManyCompare'):
            p = prefix + 'howManyCompare.' + str(howManyIndex) + '.'
            d[p + 'answer'] = int(v['answer'])
            d[p + 'data'] = v['data']
            d[p + 'approx'] = approx[v['data']]
            d[p + 'precise'] = precise[v['data']]
            d[p + 'confidence'] = v['confidence']
            howManyIndex += 1
            
        if (t == 'DidYouNotice'):
            p = prefix + 'didYouNotice.'
            d[p + 'answer'] = v['answer'] == 'yes'
            d[p + 'true'] = errorAirline > 1
            d[p + 'confidence'] = v['confidence']
            
        if (t == 'SelectAll'):
            v = deepcopy(v)
            p = prefix + 'SelectAll.'
            d[p + 'confidence'] = v['confidence']
            
            del v['confidence']
            del v['type']
            
            answers = v.keys()
            prec = [k for k,v in precise.items() if v]
            appr = [k for k,v in approx.items() if v]
            d[p + 'answer'] = ','.join(answers)
            d[p + 'approx'] = ','.join(appr)
            d[p + 'precise'] = ','.join(prec)
            d[p + 'jaccard_answer_precise'] = jaccard_similarity(answers, prec)
            d[p + 'jaccard_approx_precise'] = jaccard_similarity(appr, prec)
            d[p + 'jaccard_approx_answer'] = jaccard_similarity(appr, answers)
            
    howManyIndex = 0
    for k, v in item['preciseVis_states'].items():
        if k == 'done':
            continue
        t = v['type']
        
        if not 'confidence' in v:
            raise Exception(user, v)
        
        precise = states["{}_{}".format(1, 1)]
        approx = states["{}_{}".format(errorStates, seqStates)]
        
        prefix = 'precise.states.'
        
        if (t == 'HowMany'):
            p = prefix + 'howMany.'
            d[p + 'answer'] = int(v['answer'])
            d[p + 'approx'] = approx[focusState]
            d[p + 'precise'] = precise[focusState]
            d[p + 'confidence'] = v['confidence']
            
        if (t == 'HowManyCompare'):
            p = prefix + 'howManyCompare.' + str(howManyIndex) + '.'
            d[p + 'answer'] = int(v['answer'])
            d[p + 'data'] = v['data']
            d[p + 'approx'] = approx[v['data']]
            d[p + 'precise'] = precise[v['data']]
            d[p + 'confidence'] = v['confidence']
            howManyIndex += 1
            
        if (t == 'DidYouNotice'):
            p = prefix + 'didYouNotice.'
            d[p + 'answer'] = v['answer'] == 'yes'
            d[p + 'true'] = errorStates > 1
            d[p + 'confidence'] = v['confidence']
            
        if (t == 'SelectAll'):
            v = deepcopy(v)
            p = prefix + 'SelectAll.'
            d[p + 'confidence'] = v['confidence']
            
            del v['confidence']
            del v['type']
            
            answers = v.keys()
            prec = [k for k,v in precise.items() if v]
            appr = [k for k,v in approx.items() if v]
            d[p + 'answer'] = ','.join(answers)
            d[p + 'approx'] = ','.join(appr)
            d[p + 'precise'] = ','.join(prec)
            d[p + 'jaccard_answer_precise'] = jaccard_similarity(answers, prec)
            d[p + 'jaccard_approx_precise'] = jaccard_similarity(appr, prec)
            d[p + 'jaccard_approx_answer'] = jaccard_similarity(appr, answers)
    
    for k, v in item['timing'].items():
         d['timing.' + k] = v
            
    for k, v in item['demographics'].items():
         d['demographics.' + k] = v
    
    rows.append(d)

pprint(rows[0])



    


















    






ignoring 26364697
ignoring 53499639
ignoring 48818156
ignoring 96211827
OrderedDict([('id', '7321301'),
             ('userAgent',
              'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, '
              'like Gecko) Chrome/54.0.2840.99 Safari/537.36'),
             ('isMobile', False),
             ('studyCondition', 'difference'),
             ('firstCondition', 'states'),
             ('seqAirline', 12),
             ('errorStates', 1),
             ('seqStates', 1),
             ('focusState', 'MA'),
             ('focusAirline', 'DL'),
             ('secondCondition', 'airline'),
             ('errorAirline', 100),
             ('approx.airline.howManyCompare.0.answer', 2500),
             ('approx.airline.howManyCompare.0.data', 'B6'),
             ('approx.airline.howManyCompare.0.approx', 4700),
             ('approx.airline.howManyCompare.0.confidence', 'weak_disagree'),
             ('approx.airline.howMany.answer', 5000),
             ('approx.airline.howMany.approx', 5000),
             ('approx.airline.howMany.confidence', 'agree'),
             ('approx.airline.howManyCompare.1.answer', 3300),
             ('approx.airline.howManyCompare.1.data', 'EV'),
             ('approx.airline.howManyCompare.1.approx', 1600),
             ('approx.airline.howManyCompare.1.confidence', 'strong_agree'),
             ('approx.states.howManyCompare.0.answer', 1400),
             ('approx.states.howManyCompare.0.data', 'PR'),
             ('approx.states.howManyCompare.0.approx', 950),
             ('approx.states.howManyCompare.0.confidence', 'disagree'),
             ('approx.states.howMany.answer', 3500),
             ('approx.states.howMany.approx', 3928),
             ('approx.states.howMany.confidence', 'agree'),
             ('approx.states.howManyCompare.1.answer', -1500),
             ('approx.states.howManyCompare.1.data', 'FL'),
             ('approx.states.howManyCompare.1.approx', 4986),
             ('approx.states.howManyCompare.1.confidence', 'agree'),
             ('precise.airline.didYouNotice.answer', True),
             ('precise.airline.didYouNotice.true', True),
             ('precise.airline.didYouNotice.confidence', 'agree'),
             ('precise.airline.howManyCompare.0.answer', 200),
             ('precise.airline.howManyCompare.0.data', 'AA'),
             ('precise.airline.howManyCompare.0.approx', 5300),
             ('precise.airline.howManyCompare.0.precise', 4088),
             ('precise.airline.howManyCompare.0.confidence', 'strong_disagree'),
             ('precise.airline.SelectAll.confidence', 'strong_agree'),
             ('precise.airline.SelectAll.answer', 'UA,AA,DL,EV,B6'),
             ('precise.airline.SelectAll.approx', 'UA,AA,DL,EV,B6,WN'),
             ('precise.airline.SelectAll.precise', 'UA,AA,VX,DL,EV,B6,WN'),
             ('precise.airline.SelectAll.jaccard_answer_precise',
              0.7142857142857143),
             ('precise.airline.SelectAll.jaccard_approx_precise',
              0.8571428571428571),
             ('precise.airline.SelectAll.jaccard_approx_answer',
              0.8333333333333334),
             ('precise.airline.howManyCompare.1.answer', 1200),
             ('precise.airline.howManyCompare.1.data', 'UA'),
             ('precise.airline.howManyCompare.1.approx', 800),
             ('precise.airline.howManyCompare.1.precise', 583),
             ('precise.airline.howManyCompare.1.confidence', 'weak_agree'),
             ('precise.airline.howMany.answer', 2800),
             ('precise.airline.howMany.approx', 5000),
             ('precise.airline.howMany.precise', 5163),
             ('precise.airline.howMany.confidence', 'weak_agree'),
             ('precise.airline.howManyCompare.2.answer', 1000),
             ('precise.airline.howManyCompare.2.data', 'B6'),
             ('precise.airline.howManyCompare.2.approx', 4700),
             ('precise.airline.howManyCompare.2.precise', 5113),
             ('precise.airline.howManyCompare.2.confidence', 'agree'),
             ('precise.states.SelectAll.confidence', 'agree'),
             ('precise.states.SelectAll.answer', 'VA,MA,NY,NJ,PR'),
             ('precise.states.SelectAll.approx', 'CA,FL,VA,MA,PR,NJ,NY'),
             ('precise.states.SelectAll.precise', 'CA,FL,VA,MA,PR,NJ,NY'),
             ('precise.states.SelectAll.jaccard_answer_precise',
              0.7142857142857143),
             ('precise.states.SelectAll.jaccard_approx_precise', 1.0),
             ('precise.states.SelectAll.jaccard_approx_answer',
              0.7142857142857143),
             ('precise.states.didYouNotice.answer', True),
             ('precise.states.didYouNotice.true', False),
             ('precise.states.didYouNotice.confidence', 'strong_agree'),
             ('precise.states.howMany.answer', 3400),
             ('precise.states.howMany.approx', 3928),
             ('precise.states.howMany.precise', 3928),
             ('precise.states.howMany.confidence', 'weak_agree'),
             ('precise.states.howManyCompare.0.answer', 3500),
             ('precise.states.howManyCompare.0.data', 'CA'),
             ('precise.states.howManyCompare.0.approx', 1887),
             ('precise.states.howManyCompare.0.precise', 1887),
             ('precise.states.howManyCompare.0.confidence', 'disagree'),
             ('precise.states.howManyCompare.1.answer', 2000),
             ('precise.states.howManyCompare.1.data', 'PR'),
             ('precise.states.howManyCompare.1.approx', 950),
             ('precise.states.howManyCompare.1.precise', 950),
             ('precise.states.howManyCompare.1.confidence', 'agree'),
             ('precise.states.howManyCompare.2.answer', 200),
             ('precise.states.howManyCompare.2.data', 'VA'),
             ('precise.states.howManyCompare.2.approx', 961),
             ('precise.states.howManyCompare.2.precise', 961),
             ('precise.states.howManyCompare.2.confidence', 'strong_disagree'),
             ('timing.approximate_questions_first_html', 25798),
             ('timing.approximate_second_html', 13724),
             ('timing.precise_questions_html', 111531),
             ('timing.approximate_questions_second_html', 25278),
             ('timing.precise_instructions_html', 3750),
             ('timing.demographics_html', 28502),
             ('timing.precise_first_html', 31174),
             ('timing.approximate_first_html', 22439),
             ('timing.instructions_html', 12449),
             ('timing.index_html', 5357),
             ('timing.precise_second_html', 19730),
             ('timing.thanks_html', 1154967),
             ('demographics.problems', '-'),
             ('demographics.troubles', 'no'),
             ('demographics.improved', '-'),
             ('demographics.gender', 'Male'),
             ('demographics.age', '28'),
             ('demographics.familiar', 'some'),
             ('demographics.cheat', 'no')])

















In [25]:



    


import csv

with open('{}.csv'.format(name), 'w') as f:
    #global fieldnames
    #fieldnames = rows[0].keys()
    writer = csv.DictWriter(f, fieldnames=fieldnames)
    writer.writeheader()
    
    for row in rows:
        writer.writerow(row)

import pandas as pd

pd.read_csv('{}.csv'.format(name))



    


















    
Out[25]:










  
    

      
      id
      userAgent
      isMobile
      studyCondition
      firstCondition
      seqAirline
      errorStates
      seqStates
      focusState
      focusAirline
      ...
      timing.index_html
      timing.precise_second_html
      timing.thanks_html
      demographics.problems
      demographics.troubles
      demographics.improved
      demographics.gender
      demographics.age
      demographics.familiar
      demographics.cheat
    

  
  
    

      0
      7321301
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      difference
      states
      12
      1
      1
      MA
      DL
      ...
      5357.0
      19730
      1154967
      -
      no
      -
      Male
      28.0
      some
      no
    

    

      1
      97626871
      Mozilla/5.0 (Windows NT 6.3; WOW64) AppleWebKi...
      False
      onlyNew
      airline
      11
      100
      7
      MA
      DL
      ...
      7220.0
      14925
      15627
      None
      no
      The study was very straightforward and easy to...
      Male
      21.0
      expert
      no
    

    

      2
      29089701
      Mozilla/5.0 (X11; Linux x86_64; rv:50.0) Gecko...
      False
      both
      states
      2
      500
      13
      NY
      DL
      ...
      4081.0
      27441
      12584
      No.
      no
      No.
      male
      30.0
      familiar
      no
    

    

      3
      76395409
      Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7....
      False
      both
      states
      14
      200
      20
      MA
      B6
      ...
      8495.0
      84788
      5408
      No
      no
      No
      Female
      37.0
      familiar
      yes
    

    

      4
      9782123
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      onlyNew
      airline
      7
      100
      12
      MA
      B6
      ...
      3249.0
      17243
      10516
      No Problems
      no
      No suggestions
      male
      30.0
      some
      no
    

    

      5
      28428523
      Mozilla/5.0 (Windows NT 10.0; Win64; x64) Appl...
      False
      difference
      states
      3
      500
      16
      FL
      B6
      ...
      5174.0
      20541
      16204
      none
      no
      none
      male
      31.0
      none
      no
    

    

      6
      29657674
      Mozilla/5.0 (Windows NT 10.0; Win64; x64) Appl...
      False
      difference
      states
      7
      1
      1
      FL
      B6
      ...
      59105.0
      34038
      8755
      No
      no
      Not sure how to improve it, but looking at the...
      Male
      33.0
      familiar
      no
    

    

      7
      45694489
      Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKi...
      False
      difference
      airline
      4
      1000
      10
      FL
      AA
      ...
      44786.0
      45099
      33973
      No, not at all.
      no
      Very well-designed study. Great!
      Male
      24.0
      familiar
      no
    

    

      8
      79273300
      Mozilla/5.0 (Windows NT 6.3; WOW64) AppleWebKi...
      False
      onlyNew
      airline
      1
      1000
      18
      MA
      AA
      ...
      511301.0
      11370
      14835
      -
      no
      -
      female
      39.0
      expert
      no
    

    

      9
      43548417
      Mozilla/5.0 (Windows NT 10.0; Win64; x64) Appl...
      False
      difference
      states
      14
      100
      18
      MA
      B6
      ...
      NaN
      19665
      12410
      none
      no
      none that I can think of,
      Male
      30.0
      familiar
      no
    

    

      10
      79544988
      Mozilla/5.0 (Windows NT 6.1; Win64; x64) Apple...
      False
      both
      airline
      15
      200
      14
      MA
      B6
      ...
      5109.0
      11539
      29816
      none
      no
      -
      female
      42.0
      some
      no
    

    

      11
      20505415
      Mozilla/5.0 (Windows NT 6.1; Win64; x64) Apple...
      False
      difference
      states
      10
      200
      10
      FL
      B6
      ...
      5072.0
      22359
      6036
      NaN
      NaN
      NaN
      male
      NaN
      NaN
      NaN
    

    

      12
      88825154
      Mozilla/5.0 (Windows NT 10.0; WOW64; rv:50.0) ...
      False
      onlyNew
      airline
      11
      500
      2
      NY
      AA
      ...
      2532.0
      20983
      2005
      No problems
      no
      No suggestions, seems like a good study.
      Male
      52.0
      some
      no
    

    

      13
      21045111
      Mozilla/5.0 (Windows NT 6.1; Win64; x64) Apple...
      False
      difference
      states
      5
      200
      4
      NY
      AA
      ...
      9493.0
      27189
      8711
      -
      no
      -
      Male
      30.0
      some
      no
    

    

      14
      57315921
      Mozilla/5.0 (Windows NT 10.0; Win64; x64) Appl...
      False
      difference
      airline
      20
      200
      10
      MA
      DL
      ...
      12481.0
      77270
      32863
      No
      no
      No
      Female
      43.0
      some
      yes
    

    

      15
      71996586
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      both
      airline
      14
      500
      8
      MA
      AA
      ...
      7564.0
      12922
      15943
      No
      no
      No it was fine as is.
      Male
      42.0
      some
      no
    

    

      16
      83205605
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      both
      states
      12
      1
      1
      MA
      AA
      ...
      35203.0
      13886
      267539
      no
      no
      no
      male
      26.0
      some
      no
    

    

      17
      61154389
      Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_6...
      False
      difference
      airline
      12
      1000
      7
      FL
      DL
      ...
      2440.0
      38892
      13159
      no
      no
      no
      Female
      26.0
      familiar
      no
    

    

      18
      71607423
      Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_1...
      False
      both
      states
      1
      200
      4
      NY
      DL
      ...
      5195.0
      12148
      20278
      -
      no
      -
      male
      26.0
      some
      no
    

    

      19
      77642132
      Mozilla/5.0 (Windows NT 10.0; Win64; x64) Appl...
      False
      onlyNew
      states
      11
      100
      15
      MA
      DL
      ...
      24997.0
      38721
      69782
      No problems.
      no
      Nothing comes to mind.
      male
      40.0
      some
      no
    

    

      20
      62759895
      Mozilla/5.0 (Windows NT 10.0; rv:50.0) Gecko/2...
      False
      both
      states
      1
      1
      1
      NY
      AA
      ...
      30891.0
      12445
      12275
      no
      no
      no
      male
      31.0
      familiar
      no
    

    

      21
      21418137
      Mozilla/5.0 (Windows NT 6.3; Win64; x64) Apple...
      False
      difference
      airline
      4
      500
      6
      NY
      AA
      ...
      52003.0
      11919
      13887
      no technical problems
      no
      -
      male
      28.0
      none
      no
    

    

      22
      76024785
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      onlyNew
      airline
      3
      100
      20
      MA
      B6
      ...
      145667.0
      62106
      15231
      No
      no
      Was fine
      Male
      46.0
      familiar
      no
    

    

      23
      45840218
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      difference
      states
      1
      1
      1
      MA
      DL
      ...
      27924.0
      77830
      41253
      no
      no
      no
      female
      43.0
      familiar
      yes
    

    

      24
      95378753
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      difference
      states
      1
      200
      3
      MA
      DL
      ...
      4306.0
      13149
      23901
      No
      no
      No
      Female
      33.0
      expert
      no
    

    

      25
      26308857
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      both
      airline
      12
      200
      19
      FL
      AA
      ...
      11361.0
      26819
      26505
      none
      no
      All good.
      Male
      46.0
      familiar
      no
    

    

      26
      89431160
      Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKi...
      False
      difference
      states
      11
      1
      1
      FL
      B6
      ...
      4084.0
      27884
      10095
      I had no problems.
      no
      It seemed good overall.
      male
      28.0
      expert
      no
    

    

      27
      72925790
      Mozilla/5.0 (Windows NT 6.3; WOW64; rv:50.0) G...
      False
      difference
      airline
      3
      1
      1
      FL
      AA
      ...
      3063.0
      14258
      6309
      NaN
      no
      NaN
      Female
      35.0
      familiar
      NaN
    

    

      28
      79707421
      Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebK...
      False
      difference
      airline
      4
      200
      13
      FL
      B6
      ...
      5420.0
      26858
      27626
      no
      yes
      -
      feamle
      29.0
      some
      no
    

    

      29
      73348785
      Mozilla/5.0 (Windows NT 10.0; WOW64; rv:50.0) ...
      False
      onlyNew
      states
      5
      1000
      2
      FL
      B6
      ...
      8705.0
      31597
      16437
      none
      no
      NaN
      male
      41.0
      none
      no
    

    

      30
      77955785
      Mozilla/5.0 (Windows NT 6.1; Win64; x64) Apple...
      False
      difference
      states
      9
      500
      9
      MA
      AA
      ...
      NaN
      18487
      62111
      None.
      no
      Maybe improve on the scale to make estimations...
      Male
      24.0
      some
      no
    

  



31 rows × 111 columns



















In [ ]:

Content source: domoritz/bias-study

Similar notebooks:

notebook.community | gallery | about