def import_imdb_multi_graph(self):
if os.path.exists(self.graph_path):
multi_gnx = nx.read_gpickle(self.graph_path)
else:
from IMDb_data_preparation import main
multi_gnx = main()
nx.write_gpickle(multi_gnx, 'pkl/IMDb_multi_graph1.gpickle')
return multi_gnx
def import_imdb_multi_graph(self):
path = os.path.join(self.data_name, 'IMDb_multi_graph.gpickle')
if os.path.exists(path):
multi_gnx = nx.read_gpickle(path)
else:
from IMDb_data_preparation import main
multi_gnx = main()
nx.write_gpickle(multi_gnx, path)
return multi_gnx
# dict_event2vec_embeddings = embedding_model.create_event2vec_embeddings()
# nodes = list(dict_event2vec_embeddings.keys())
# relevant_edges = edges_to_predict(multi_graph)
# true_edges = choose_true_edges(relevant_edges, number)
# false_edges = choose_false_edges(multi_graph, relevant_edges, number)
true_edges = choose_true_edges(graph, number)
false_edges = choose_false_edges(graph, number, args.data_name, True)
my_dict, X, Y = calculate_classifier_value(dict_embeddings, true_edges,
false_edges, number)
X_train, X_test, Y_train, Y_test = sk_ms.train_test_split(
X, Y, test_size=ratio_arr[4])
prediction, probs = predict_edge_classification(X_train, X_test, Y_train,
Y_test)
micro, macro, acc, auc = evaluate_edge_classification(prediction, Y_test)
print('acc: ', acc)
print('auc: ', auc)
print('micro: ', micro)
print('macro: ', macro)
# micro, macro, acc, auc = exp_lp(X, Y, ratio_arr, 3)
# avg_micro, avg_macro, avg_acc, avg_auc = calculate_all_avg_scores(micro, macro, acc, auc, 3)
# all_micro.append(avg_micro)
# all_macro.append(avg_macro)
# all_acc.append(avg_acc)
# all_auc.append(avg_auc)
# fig1, fig2, fig3, fig4 = split_vs_score(all_micro[0], all_macro[0], all_micro[1], all_macro[1], all_acc[0],
# all_acc[1], all_auc[0], all_auc[1], ratio_arr)
# plt.show()
main()
