def cv10_ensemble(known_dataset, known_targets, known_dataset_scaled, dt, knn, svm, fusion_algorithm, ids, prt=False, file_name=None): error_rates = 0 hp_rates = 0 hr_rates = 0 hf_rates = 0 cp_rates = 0 cr_rates = 0 cf_rates = 0 for i in range(NR_FOLDS): error, hp, hr, hf, cp, cr, cf = cross_validation_ensemble(known_dataset, known_targets, known_dataset_scaled, dt, knn, svm, fusion_algorithm, ids, prt, file_name) error_rates += error hp_rates += hp hr_rates += hr hf_rates += hf cp_rates += cp cr_rates += cr cf_rates += cf if prt and (float(error_rates) / NR_FOLDS) <= 0.4: save_output(file_name, error_rates, hp_rates, hr_rates, hf_rates, cp_rates, cr_rates, cf_rates, NR_FOLDS) print 'Final error %f' % (float(error_rates) / NR_FOLDS) print 'Final accuracy %f' % (1 - (float(error_rates) / NR_FOLDS)) print 'Highval precision %f' % (float(hp_rates) / NR_FOLDS) print 'Highval recall %f' % (float(hr_rates) / NR_FOLDS) print 'Highval f1 %f' % (float(hf_rates) / NR_FOLDS) print 'Civil precision %f' % (float(cp_rates) / NR_FOLDS) print 'Civil recall %f' % (float(cr_rates) / NR_FOLDS) print 'Civil f1 %f' % (float(cf_rates) / NR_FOLDS)
def main():
'''User-Inputted Recipie Title'''
#url = generateURL()
'''Hard Coded URLs'''
# url = "http://allrecipes.com/recipe/brown-rice-and-quinoa-sushi-rolls/"
# url = "http://allrecipes.com/recipe/Boilermaker-Tailgate-Chili/"
#url = "http://allrecipes.com/recipe/jerk-chicken/"
#link = urllib.urlopen(url)
#page = link.read()
'''Local Cached Webpages'''
# url = "../data/Burger"
# url = "../data/Cake"
url = "../data/Stir-Fry"
f = open(url + ".html")
page = f.read()
if len(sys.argv) < 2:
recipe = parse_recipe(page)
recipe = unabridgeMeasure(recipe)
prettyPrintRecipe(recipe)
save_output(url, recipe)
elif len(sys.argv) == 2:
case_num = int(sys.argv[1])
if case_num < 1 or case_num > 8:
print('You can only pick from [1-8]')
return
recipe = parse_recipe(page)
recipe = unabridgeMeasure(recipe)
kb = KnowledgeBase()
if case_num == 1:
tf_recipe = kb.transform_cuisine("italian", recipe)
elif case_num == 2:
tf_recipe = kb.transform_cuisine("chinese", recipe)
elif case_num == 3:
tf_recipe = kb.transform_diet("vegetarian", recipe)
elif case_num == 4:
tf_recipe = kb.transform_diet("pescatarian", recipe)
elif case_num == 5:
tf_recipe = kb.transform_healthy("low-fat", recipe)
elif case_num == 6:
tf_recipe = kb.transform_healthy("low-sodium", recipe)
elif case_num == 7:
tf_recipe = transformQty(2, recipe)
elif case_num == 8:
tf_recipe = transformQty(3, recipe)
prettyPrintRecipe(tf_recipe)
save_output(url, tf_recipe)
else:
print(
'Too many arguments. You can either either just call main.py to see the recipe or pass in a single integer to select a transformation.'
)
print 'Highval f1 %f' % hf
print 'Civil precision %f' % cp
print 'Civil recall %f' % cr
print 'Civil f1 %f' % cf
return error_rate, f1, model, (hp, hr, hf), (cp, cr, cf)
if __name__ == "__main__":
training_spreadsheet = Spreadsheet(project_data_file)
training_data = Data(training_spreadsheet)
training_targets = training_data.targets
testing_spreadsheet = Spreadsheet(addendum_data_file, upsampling=False)
testing_data = Data(testing_spreadsheet, upsampling=False)
testing_targets = testing_data.targets
[training_data, features] = parse_theme('all')
[testing_data, feats] = parse_theme_from_file('all', addendum_data_file)
assert features == feats
tech = raw_input("Enter algorithm. Choose between lr, dt, knn, svm")
file_name = "new_single_" + tech + ".txt"
for i in range(100):
error_rate, f1, model, (hp, hr, hf), (cp, cr, cf) = new_data_single(training_data, training_targets, testing_data, testing_targets, tech)
save_output(file_name, error_rate, hp, hr, hf, cp, cr, cf, 1)
def generate_models():
'''
This function generates 26 neural network models; 13 for each type of wine:
red and white. 11 of those models utilize PCA data, 1 is trained on all raw
feature data, and 1 is trained on 4 sensory features to equal 13 per type.
'''
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
filepath_red = './raw_data/winequality-red.csv'
filepath_white = './raw_data/winequality-white.csv'
red_wine_df = di(filepath_red)
red_pc_dict = pc_analysis(red_wine_df)
red_pc_model_list = nn(red_pc_dict)
red_raw_quality = pd.DataFrame(red_wine_df['quality'])
red_raw_quality.columns = ['quality']
del red_wine_df['quality']
red_raw_data_dict = {
'11 Raw Data - Red': [None, red_wine_df, red_raw_quality]
}
red_raw_data_model = nn(red_raw_data_dict)
red_4_columns = pd.DataFrame(
(red_wine_df['fixed acidity'], red_wine_df['citric acid'],
red_wine_df['residual sugar'], red_wine_df['alcohol'])).transpose()
red_4_columns_dict = {
'4 Columns Raw Data - Red': [None, red_4_columns, red_raw_quality]
}
red_4_columns_model = nn(red_4_columns_dict)
red_model_list = (red_pc_model_list + red_raw_data_model +
red_4_columns_model)
save_output(red_model_list, 'red_performance.txt', red_pc_dict)
white_wine_df = di(filepath_white)
white_pc_dict = pc_analysis(white_wine_df)
white_pc_model_list = nn(white_pc_dict)
white_raw_quality = pd.DataFrame(white_wine_df['quality'])
white_raw_quality.columns = ['quality']
del white_wine_df['quality']
white_raw_data_dict = {
'11 Raw Data - White': [None, white_wine_df, white_raw_quality]
}
white_raw_data_model = nn(white_raw_data_dict)
white_4_columns = pd.DataFrame(
(white_wine_df['fixed acidity'], white_wine_df['citric acid'],
white_wine_df['residual sugar'],
white_wine_df['alcohol'])).transpose()
white_4_columns_dict = {
'4 Columns Raw Data - White':
[None, white_4_columns, white_raw_quality]
}
white_4_columns_model = nn(white_4_columns_dict)
white_model_list = (white_pc_model_list + white_raw_data_model +
white_4_columns_model)
save_output(white_model_list, 'white_performance.txt', white_pc_dict)