ideatory.py

import pandas as pd
import numpy as np

#attendance_data = pd.read_csv('AttendanceData.csv',index_col=2,
#                              parse_dates=True)
#attendance_data.columns = ['event_id', 'user_id']

event_data = pd.read_csv('EventData.csv',#skipfooter=1,
                          index_col=0)
event_data.columns = ['latitude', 'longitude']
event_data.dropna(inplace=True)

user_data = pd.read_csv('UserData.csv', #skipfooter=1,
                        index_col=0)
user_data.columns = ['latitude', 'longitude']
user_data.dropna(inplace=True)

training_data = pd.read_csv('ratings_for_train.csv') 
                            #skipfooter=1)
training_data.columns = ['user_id','event_id','rating']
training_data.dropna(inplace=True)

#testing_data = pd.read_csv('ratings_for_eval.csv',usecols=['User ID', 'Event ID'])
#testing_data.columns = ['user_id','event_id']

def clean_missing(training_data, user_data, event_data):
    """
    Drop training examples for users and events whose data isn't available.
    """
    users_in_training = training_data['user_id'].unique()
    events_in_training = training_data['event_id'].unique()
    for uid in users_in_training:
        if uid not in user_data.index:
            inds = training_data[training_data['user_id'] == uid].index
            training_data.drop(inds, inplace=True)
    for eid in events_in_training:
        if eid not in event_data.index:
            inds = training_data[training_data['event_id'] == eid].index
            training_data.drop(inds, inplace=True)
    return training_data, user_data, event_data

#def fill_coordinates(training_data, user_data, event_data):
#    """
#    Add latitude and longitude columns to the training data containing the
#    coordinates for each row.
#    """
#    training_data['ulat'] = np.zeros((training_data.shape[0],))
#    training_data['ulong'] = np.zeros((training_data.shape[0],))
#    training_data['elat'] = np.zeros((training_data.shape[0],))
#    training_data['elong'] = np.zeros((training_data.shape[0],))
#    event_ids = event_data.index.unique()
#    user_ids = user_data.index.unique()
#    for uid in user_ids:
#        ulat, ulong = user_data.ix[uid]