def create_matrix(states):
# Create item-centric preference matrix from DataFrame
itemPrefs = {}
for idx, row in states.iterrows():
itemPrefs[row["story_id"]] = {}
for st in parameters.state_names():
itemPrefs[row["story_id"]][st] = row[st]
return itemPrefs
def create_states_dataframe(input_data):
# Create DataFrame for States
data = []
for row in input_data:
d = json.loads(row)
story_id = d['data']['story_id']
# Create State Vector
statedata = {}
for e in d['data']['citiesbycountry']:
state = e[0][3:5]
if state in statedata:
statedata[state] += e[1]
else:
statedata[state] = e[1]
statevector = []
for state in parameters.state_names():
if state in statedata:
statevector.append(statedata[state])
else:
statevector.append(0)
new_row = [story_id] + statevector
data.append(new_row)
states_dataframe = pandas.DataFrame(data, columns = ['story_id'] + parameters.state_names())
return states_dataframe
# Retrieve linkinfo
sql_query = "SELECT * FROM " + linkinfo_tablename
cursor.execute(sql_query)
rows = cursor.fetchall()
data = tuple_to_list(rows)
linkinfo = pandas.DataFrame(data, columns = ['old_ix', 'story_id', 'rate', 'thumbnail_url', 'description', 'title', 'url', 'clicks'])
# Retrieve state data
sql_query = "SELECT * FROM " + states_tablename
cursor.execute(sql_query)
rows = cursor.fetchall()
data = tuple_to_list(rows)
states = pandas.DataFrame(data, columns = ['old_ix', 'story_id'] + parameters.state_names())
# Retrieve country data
sql_query = "SELECT * FROM " + countries_tablename
cursor.execute(sql_query)
rows = cursor.fetchall()
data = tuple_to_list(rows)
countries = pandas.DataFrame(data, columns = ['old_ix', 'story_id', 'count', 'us'])
info = {}
for ix, row in linkinfo.iterrows():
story_id = row['story_id']
info[story_id] ={}
info[story_id]['title'] = row['title']
info[story_id]['description'] = row['description']