import pandas as pd
from unipath import Path
from __init__ import unfold, survey_dir, csv_output_dir
responses = pd.read_json(Path(survey_dir, 'responses.json'))
for response_field in ['selection', 'question']:
responses = unfold(responses, response_field)
del responses['fields']
responses = responses[['selection', 'question']]
responses.rename(columns={'question': 'question_id'}, inplace=True)
questions = pd.read_csv(Path(survey_dir, 'questions.csv'))
responses = responses.merge(questions)
responses = responses[[
'survey_id',
'survey_label',
'question_id',
'message_id',
'generation',
'game_name',
'chain_name',
'answer',
'selection',
]]
import pandas as pd
from unipath import Path
from __init__ import unfold, survey_dir, csv_output_dir
responses = pd.read_json(Path(survey_dir, 'responses.json'))
for response_field in ['selection', 'question']:
responses = unfold(responses, response_field)
del responses['fields']
responses = responses[['selection', 'question']]
responses.rename(columns={'question': 'question_id'}, inplace=True)
questions = pd.read_csv(Path(survey_dir, 'questions.csv'))
responses = responses.merge(questions)
responses = responses[[
'survey_id', 'survey_label',
'question_id', 'message_id', 'generation', 'game_name', 'chain_name',
'answer', 'selection',
]]
responses.to_csv(Path(csv_output_dir, 'responses.csv'), index=False)
import pandas as pd
from unipath import Path
from __init__ import unfold, survey_dir
messages = pd.read_json(Path(survey_dir, 'messages.json'))
del messages['model']
for message_field in [
'generation', 'num_children', 'audio', 'chain', 'parent'
]:
messages = unfold(messages, message_field)
del messages['fields']
def extract_from_path(frame, path_col, name, index):
frame[name] = frame[path_col].str.split('/').str.get(index)
return frame
for i, name in enumerate(['game_name', 'chain_name', 'message_name']):
messages = extract_from_path(messages, 'audio', name, i)
messages = messages.ix[messages.game_name != 'test-game']
messages = messages.sort(['game_name', 'chain_name', 'message_name'])
messages = messages.rename(columns={
'pk': 'message_id',
import pandas as pd
from unipath import Path
from __init__ import unfold, survey_dir
questions = pd.read_json(Path(survey_dir, 'questions.json'))
for question_field in ['choices', 'given', 'survey', 'answer']:
questions = unfold(questions, question_field)
del questions['fields']
del questions['model']
del questions['choices']
questions.rename(columns={'pk': 'question_id', 'survey': 'survey_id', 'given': 'message_id'}, inplace=True)
survey_info = pd.DataFrame({
'survey_id': [1, 2, 3, 4, 12, 13],
'survey_label': ['between', 'within', 'between', 'within', 'between-splish', 'within-splish']
})
questions = questions.merge(survey_info)
messages = pd.read_csv(Path(survey_dir, 'messages.csv'))
def pick_ancestor(message):
if message in [1,2,3,4] + [138,139,140,141]:
return message
else:
parent = messages.ix[messages.message_id == message, 'parent_id']
return pick_ancestor(int(parent))
import pandas as pd
from unipath import Path
from __init__ import unfold, survey_dir
messages = pd.read_json(Path(survey_dir, 'messages.json'))
del messages['model']
for message_field in ['generation', 'num_children', 'audio', 'chain', 'parent']:
messages = unfold(messages, message_field)
del messages['fields']
def extract_from_path(frame, path_col, name, index):
frame[name] = frame[path_col].str.split('/').str.get(index)
return frame
for i, name in enumerate(['game_name', 'chain_name', 'message_name']):
messages = extract_from_path(messages, 'audio', name, i)
messages = messages.ix[messages.game_name != 'test-game']
messages = messages.sort(['game_name', 'chain_name', 'message_name'])
messages = messages.rename(columns={'pk': 'message_id', 'chain': 'chain_id', 'parent': 'parent_id'})
messages.to_csv(Path(survey_dir, 'messages.csv'), index=False)
