print("Hidden Lyers ", HIDDEN_LAYERS) print("PAST_DAYS ", PAST_DAYS) print("TESTING_PERCENTAGE ", TESTING_PERCENTAGE) print("Epochs Num ", n_epochs) print("Normalization ", normalization) Experiment_KEY = 'Hidden_' + str(HIDDEN_LAYERS) + '_DAYS_' + str( PAST_DAYS) + '_nepoch_' + str(n_epochs) + '_normalization_' + str( normalization) + '_features_' + str(n_in) print(Experiment_KEY) cluster_subset = [1, 5, 11, 19] airports_clusters = get_clusters() tst_indexes = get_test_indexes() airports, airports_data = parse_data(normalization=normalization) real_test_data = get_real_test_target(tst_indexes) airports_clusters_data = {} airports_all_data = { 'features_trn': [], 'targets_trn': [], 'features_tst': [], 'targets_tst': [], 'targets_tst_real': [] } for air_id in airports: cluster_id = airports_clusters[air_id] if cluster_id in cluster_subset: data = [] for rec in airports_data[air_id]: data.append(rec)
n_in = 18 n_out = 2 # normalization='zcore' print("PAST_DAYS ", PAST_DAYS) print("TESTING_PERCENTAGE ", TESTING_PERCENTAGE) print("Normalization ", normalization) cluster_subset=[1,5,11,19] airports_clusters = get_clusters() err_counter = 0 err_mintmp = 0 err_maxtmp = 0 tst_indexes = get_test_indexes() airports, airports_data = parse_data(normalization=normalization) real_test_data=get_real_test_target(tst_indexes) airports_clusters_data = {} airports_all_data = {'features_trn': [], 'targets_trn': [], 'features_tst': [], 'targets_tst': [],'targets_tst_real':[]} air_count=0 for air_id in airports: cluster_id = airports_clusters[air_id] if cluster_id in cluster_subset: air_count+=1 data = [] for rec in airports_data[air_id]: data.append(rec) features_trn, targets_trn, features_tst, targets_tst ,targets_tst_real = divide_to_sequence(data, tst_indexes, feature_num=n_in, PAST_DAYS=PAST_DAYS,real_test_data=real_test_data)
import numpy as np from parse_data import * from download_images import * from update_prices import * from add_features import * labels = ["idd", "city", "exhibition", "artist", "title", "price", "sold", "avg_estimate", "signed", "area", \ "volume", "year_created", "auction_lot", "auction_house", "auction_date", \ "avg_log_price_sold_before", "median_price_sold_before", "num_artworks", \ "num_artists", "sale_rate_before", "img_url", "volatility_returns_before", "mean_returns_before", "skew_prices_before", "medium", "lots_per_artist", "num_artworks_ratio"] cities = ["New York", "London", "Paris"] for city in cities: print(city) parse_data(city, labels) update_prices(city)