def test_build(self): look_back_years = 3 samples_set = SamplesSet(look_back_years) samples_set.t_loc = "./foc/forecaster/tests/test_sample_selection.xls" train_samples, test_samples = samples_set.build_from_crises_file(["USA", "DEU"], ["SP.POP.65UP.TO.ZS"], 0.50) self.assertEqual(len(train_samples), 4) self.assertEqual(len(test_samples), 3)
def test_build(self): look_back_years = 3 samples_set = SamplesSet(look_back_years) samples_set.t_loc = "./foc/forecaster/tests/test_sample_selection.xls" train_samples, test_samples = samples_set.build_from_crises_file( ["USA", "DEU"], ["SP.POP.65UP.TO.ZS"], 0.50) self.assertEqual(len(train_samples), 4) self.assertEqual(len(test_samples), 3)
def run(): t1 = time() print("Configuration loaded from:") try: visualising = (sys.argv[1]=="visualise") except IndexError: visualising = False if visualising: # we'll only draw stuff print(vis_conf.__file__) visualiser = Visualiser() visualiser.draw() else: # we're gonna build a data set # conf path print(conf.__file__) # size estimation size_float = 4. B = (conf.end_date-conf.start_date+1)*len(conf.countries)*len(conf.indicators)*size_float KB = B/1000. MB = KB/1000. print("Attempting to fetch roughly %.2f KB (%.2f MB) from the World Bank." % (KB, MB)) # start work samples_set = SamplesSet(conf.look_back_years, conf.cache_enabled, conf.cache_host, conf.cache_port) samples_set.build_from_crises_file(conf.countries, conf.indicators, conf.testing_percentage) writer = Writer() separate_train_test = (conf.testing_percentage!=0.0) writer.write(samples_set, conf.output_formats, conf.output_location, separate_train_test) t2 = time() print("Done!") print("Duration:") duration = t2-t1 print("%.2f s (%.2f min)" % (duration, (duration)/60.)) if visualising: visualiser.show()