| Index: trunk/tools/editor_trends/analyses/plugins/taxonomy_burnout.py |
| — | — | @@ -18,44 +18,109 @@ |
| 19 | 19 | __date__ = '2011-01-25' |
| 20 | 20 | __version__ = '0.1' |
| 21 | 21 | |
| | 22 | +from datetime import date |
| 22 | 23 | |
| 23 | | -def taxonomy_burnout(var, editor, **kwargs): |
| 24 | | - ''' |
| | 24 | +''' |
| | 25 | + taxonomy_burnout |
| | 26 | + ================ |
| | 27 | + |
| | 28 | + This is a Taxonomy project metric. The "Editor Burnout" metric is intended to measure |
| | 29 | + editors that make a large number of edits in a single month and then make relatively few |
| | 30 | + in the following months. The exact number of edits and months are defined by the |
| | 31 | + following gloabal variables: |
| | 32 | + |
| | 33 | + ONE_MONTH_EDIT_COUNT_CUTOFF |
| | 34 | + MONTHS_AFTER_CUTOFF_TO_RECORD_BURNOUT |
| | 35 | + |
| | 36 | + While the average number of edits per month beyond MONTHS_AFTER_CUTOFF_TO_RECORD_BURNOUT |
| | 37 | + that classify an editor as a "burned out" is defined in: |
| | 38 | + |
| | 39 | + CUTOFF_RATE_FOR_BURNOUT_EDITORS |
| | 40 | + |
| | 41 | + More documentation may be found at: |
| | 42 | + |
| | 43 | + http://meta.wikimedia.org/wiki/Contribution_Taxonomy_Project/Research_Questions |
| | 44 | + |
| 25 | 45 | If you have questions about how to use this plugin, please visit: |
| 26 | 46 | http://meta.wikimedia.org/wiki/Wikilytics_Plugins |
| 27 | | - ''' |
| | 47 | + |
| | 48 | +''' |
| | 49 | +def taxonomy_burnout(var, editor, **kwargs): |
| | 50 | + |
| | 51 | + """ These parameters can be used to tune the burnout analyses """ |
| | 52 | + global MONTHS_AFTER_CUTOFF_TO_RECORD_BURNOUT |
| | 53 | + global CUTOFF_RATE_FOR_BURNOUT_EDITORS |
| | 54 | + global ONE_MONTH_EDIT_COUNT_CUTOFF |
| | 55 | + |
| | 56 | + MONTHS_AFTER_CUTOFF_TO_RECORD_BURNOUT = 6 |
| | 57 | + CUTOFF_RATE_FOR_BURNOUT_EDITORS = 10 |
| | 58 | + ONE_MONTH_EDIT_COUNT_CUTOFF = 149 |
| | 59 | + |
| | 60 | + |
| | 61 | + """ Record editor properties """ |
| 28 | 62 | new_wikipedian = editor['new_wikipedian'] |
| 29 | 63 | edits = editor['edit_count'] |
| 30 | | - cutoff = kwargs.get('cutoff', 149) |
| | 64 | + cutoff = kwargs.get('cutoff', ONE_MONTH_EDIT_COUNT_CUTOFF) |
| 31 | 65 | username = editor['username'] |
| 32 | 66 | |
| | 67 | + made_cutoff = False |
| 33 | 68 | burnout = False |
| 34 | | - sum = 0.0 |
| 35 | | - count = 0.0 |
| | 69 | + |
| | 70 | + total_edits_since_cutoff = 0.0 |
| | 71 | + total_months_since_cutoff = 0.0 |
| | 72 | + |
| | 73 | + zero_edit_months = dict() |
| 36 | 74 | |
| | 75 | + """ |
| | 76 | + new_wikipedian is False if the user is not classified as a wikipedian |
| | 77 | + and returns the date of their becoming a Wikipedian otherwise |
| | 78 | + |
| | 79 | + In that case iterate through the months to present trying to find: |
| | 80 | + (1) If they meet ONE_MONTH_EDIT_COUNT_CUTOFF |
| | 81 | + (2) If the fall into the burnout category thereafter |
| | 82 | + """ |
| | 83 | + |
| 37 | 84 | if new_wikipedian: |
| 38 | 85 | years = edits.keys() |
| | 86 | + |
| 39 | 87 | for year in years: |
| 40 | 88 | months = edits[year].keys() |
| | 89 | + |
| 41 | 90 | for month in months: |
| 42 | | - try: |
| 43 | | - if edits[year][month].get('0', 0) > cutoff: |
| 44 | | - burnout = True |
| 45 | | - if burnout == True: |
| 46 | | - n = edits[year][month].get('0', 0) |
| 47 | | - sum += n |
| 48 | | - if n > 0: |
| 49 | | - count += 1.0 |
| 50 | | - except (AttributeError, KeyError): |
| 51 | | - print 'Editor %s does not have data for year: %s and month %s. Data: %s' \ |
| 52 | | - % (username, year, month, edits[year]) |
| | 91 | + |
| | 92 | + if edits[year][month].get('0', 0) > cutoff: |
| | 93 | + made_cutoff = True |
| | 94 | + if made_cutoff == True: |
| | 95 | + |
| | 96 | + """ Count every month after the cutoff is made """ |
| | 97 | + total_months_since_cutoff += 1.0 |
| | 98 | + |
| | 99 | + """ Handle cases where no edits were recorded for a given month """ |
| | 100 | + try: |
| | 101 | + edit_count_month = edits[year][month].get('0', 0) |
| | 102 | + except (AttributeError, KeyError): |
| | 103 | + edit_count_month = 0 |
| | 104 | + |
| | 105 | + """ Record that this editor had no edits on this month """ |
| | 106 | + try: |
| | 107 | + zero_edit_months[username].append(date(year, month, 1)) |
| | 108 | + except KeyError: |
| | 109 | + zero_edit_months[username] = list() |
| | 110 | + zero_edit_months[username].append(date(year, month, 1)) |
| | 111 | + |
| | 112 | + total_edits_since_cutoff = total_edits_since_cutoff + edit_count_month |
| | 113 | + |
| | 114 | + if total_edits_since_cutoff / total_months_since_cutoff < CUTOFF_RATE_FOR_BURNOUT_EDITORS and \ |
| | 115 | + total_months_since_cutoff > MONTHS_AFTER_CUTOFF_TO_RECORD_BURNOUT and burnout == False: |
| | 116 | + burnout = True |
| | 117 | + burnout_date = date(year, month, 1) |
| | 118 | + |
| | 119 | + if burnout: |
| | 120 | + avg_edit = total_edits_since_cutoff / total_months_since_cutoff |
| 53 | 121 | |
| 54 | | - if burnout and sum / count > 10: |
| 55 | | - avg_edit = sum / count |
| 56 | | - |
| 57 | 122 | try: |
| 58 | | - var.add(new_wikipedian, avg_edit, {'username' : username}) |
| 59 | | - except Exception, error: |
| | 123 | + var.add(burnout_date, avg_edit, {'username' : username}) |
| | 124 | + except Exception as error: |
| 60 | 125 | print 'user: %s error: %s' % (username, error) |
| 61 | 126 | |
| 62 | 127 | return var |