lookalike_001.py

def dataframe_from_file(filename):
    try:
        taxi_data = pd.read_csv(filename, names=['taxi_id', 'ts', 'longitude', 'latitude'], parse_dates=['ts'])

        if len(taxi_data.index) == 0:
            print("skipping {} as empty".format(filename))
            return None
        
        taxi_data['geohash'] = np.vectorize(lambda longitude, latitude: pgh.encode(latitude, longitude, precision=6))(
            taxi_data['longitude'], 
            taxi_data['latitude']
        )

        return taxi_data[['taxi_id', 'geohash']].drop_duplicates()
    except:
        print("cannot parse {}".format(filename))
        
        return None