def run_example(stats=False):
encoder = OneHotEncoder({'animal': 2, 'color': 1}, ['weight', 'height'])
data = [{'animal': 'cat', 'color': 'blue', 'weight': 6.0, 'height': 88.9, 'extra_junk': 'blah'},
{'animal': 'cat', 'color': 'red', 'weight': 3.0, 'height': 44.9},
{'animal': 'dog', 'color': 'yellow', 'weight': 5.5, 'height': 2.5},
{'animal': 'fish', 'color': 'blue', 'weight': 7.0, 'height': 3233.2},
{'animal': 'cat', 'color': 'magenta', 'weight': 2.0, 'height': 666.6},
{'animal': 'mouse', 'color': 'red', 'weight': 0.0, 'height': 55.5},
{'animal': 'mouse', 'color': 'blah', 'weight': 99.9, 'height': 33}]
encoder.load_from_data_stream(data)
encoded_data = encoder.encode_data(data)
data_decoded = encoder.decode_data(encoded_data)
expected = [{'height': 88.9, 'weight': 6.0, 'animal': 'cat', 'color': 'blue'},
{'height': 44.9, 'weight': 3.0, 'animal': 'cat', 'color': 'UNKNOWN_CATEGORICAL_LEVEL'},
{'height': 2.5, 'weight': 5.5, 'color': 'UNKNOWN_CATEGORICAL_LEVEL', 'animal': 'UNKNOWN_CATEGORICAL_LEVEL'},
{'height': 3233.2, 'weight': 7.0, 'color': 'blue', 'animal': 'UNKNOWN_CATEGORICAL_LEVEL'},
{'height': 666.6, 'weight': 2.0, 'animal': 'cat', 'color': 'UNKNOWN_CATEGORICAL_LEVEL'},
{'height': 55.5, 'weight': 0.0, 'animal': 'mouse', 'color': 'UNKNOWN_CATEGORICAL_LEVEL'},
{'height': 33, 'weight': 99.9, 'animal': 'mouse', 'color': 'UNKNOWN_CATEGORICAL_LEVEL'}]
assert data_decoded == expected
# add number stats?
if stats:
encoder.add_numeric_stats(data)
# check the package
packaged = encoder.package_data()
return packaged
def get_round_trip_decoded(stats=False, omit_cols=None):
encoder = OneHotEncoder({'animal': 2, 'color': 1}, ['weight', 'height'], omit_cols=omit_cols)
data = [{'animal': 'cat', 'color': 'blue', 'weight': 6.0, 'height': 88.9, 'extra_junk': 'blah'},
{'animal': 'cat', 'color': 'red', 'weight': 3.0, 'height': 44.9},
{'animal': 'dog', 'color': 'yellow', 'weight': 5.5, 'height': 2.5},
{'animal': 'fish', 'color': 'blue', 'weight': 7.0, 'height': 3233.2},
{'animal': 'cat', 'color': 'magenta', 'weight': 2.0, 'height': 666.6},
{'animal': 'mouse', 'color': 'red', 'weight': 0.0, 'height': 55.5},
{'animal': 'mouse', 'color': 'blah', 'weight': 99.9, 'height': 33}]
encoder.load_from_data_stream(data)
encoded_data = encoder.encode_data(data)
data_decoded = encoder.decode_data(encoded_data)
# add number stats?
if stats:
encoder.add_numeric_stats(data)
# check the package
packaged = encoder.package_data()
return data_decoded, packaged