def make_train_set():
"""
make train set from filtered data with specified categories level and corpus size limits
"""
data = read_as_json(FILTERED_DATA_PATH)
corpora = get_n_level_categories_corpora(data, level=2)
corpora = filter_corpora_by_each_class_size(corpora, lower_limit=200, upper_limit=1000)
save_json_as_csv(corpora, CATEGORIES_CORPORA_PATH)
def print_categories():
"""
print all items categories in filtered dataset
"""
data = read_as_json(FILTERED_DATA_PATH)
categories = get_all_level_categories(data)
categories = sorted(categories)
for category in categories:
print category
