def generate_recommendations(self):
self.logging.debug('starting worker....')
worker = Worker()
rec = Recommender()
self.logging.debug("starting to process user data")
worker.process_user_data()
self.logging.debug("starting to process document data")
worker.process_document_data()
worker.get_all_tags()
worker.get_school_types()
self.logging.debug("constructing product matrix")
product_matrix = worker.build_product_matrix()
documents_data = worker.get_product_matrix_data(product_matrix)
for doc_col in documents_data:
self.logging.debug("Processing batch of {0} matrix docs".format(len(doc_col)))
# we dont want to process single item lists
if len(doc_col) > 2:
self.logging.debug("generating recommendations....")
downloaded_documents_matrix = worker.build_downloaded_document_matrix(doc_col)
df = rec.get_data_frame(downloaded_documents_matrix)
yield self.process_document_collection(df)
