def LDA_TAGS_NUM_AVERAGE_BUILDER():
pipeline_components = [(json_layer, JsonToTagsNumericalTransformer()),
(embedder, LdaTransformer()),
(tree, BallTreePredictor(average=True)),
(output, DumbOutput())]
return Pipeline(pipeline_components)
def LDA_TAGS_NUM_GREEDY_BUILDER():
pipeline_components = [(json_layer, JsonToTagsNumericalTransformer()),
(embedder, LdaTransformer()),
(tree, BallTreePredictor()),
(output, GreedyOutput())]
return Pipeline(pipeline_components)
def GLOVE_GREEDY_BUILDER():
with open("word_hash", "rb") as file:
word_hash = pickle.load(file)
pipeline_components = [
(json_layer, JsonTransform()),
(embedder, TfIdfGloveTransformer(word_embedder=word_hash)),
(tree, BallTreePredictor()), (output, GreedyOutput())
]
return Pipeline(pipeline_components)