Load a dataset and cache it for further calls

load_and_cache.py

import numpy as np
import pickle

def load_and_cache(filename):
    output = None
    cache_dir = '/tmp/my_project_cache/'
    os.makedirs(cache_dir, mode=0o744, exist_ok=True)
    cache = cache_dir + filename + '.pkl'
    if not os.path.exists(cache):
        output = np.genfromtxt(filename, dtype=float, delimiter=',')  # TODO: Fill your logic
        print('Caching {:1} into {:2}'.format(filename, cache))
        with open(cache, 'wb') as f:
            pickle.dump(output, f)
    else:
        print('Reading {:1} from cache {:2}'.format(filename, cache))
        with open(cache, 'rb') as f:
            output = pickle.load(f)
    return output