def get_paper_model(n_neighbors):
"""Build and return the paper model based on pickled kd-tree."""
print("", flush=True) # add a blank line before log section
server_log("Start initializing paper models")
pkl_filename = "data/kd_tree.pkl"
with open(pkl_filename, "rb") as fh:
kd_tree = pickle.load(fh)
paper_model = KNeighborsClassifier(n_neighbors)
paper_model.n_samples_fit_ = len(kd_tree.idx_array)
paper_model.effective_metric_ = "euclidean"
paper_model._fit_method = 'kd_tree'
paper_model._tree = kd_tree
server_log("Finished loading paper models\n")
return paper_model
def create_paper_models(n_neighbors):
"""Return a list of paper models based on pickled sub-kd-trees."""
print('', flush=True) # add a blank line before log section
server_log("Start initializing paper models")
N_PAPER_MODELS = 4 # number of sub-models for paper dataset
PAPER_MODEL_SIZE = 431539 # number of nodes in each sub-model
paper_models = list()
for id in range(1, N_PAPER_MODELS + 1):
pkl_filename = f"data/paper_dataset/sub{id}.pkl"
sub_model = KNeighborsClassifier(n_neighbors)
sub_model.n_samples_fit_ = PAPER_MODEL_SIZE
sub_model.effective_metric_ = 'euclidean'
sub_model._fit_method = 'kd_tree'
sub_model._tree = pickle.load(open(pkl_filename, "rb"))
paper_models.append(sub_model)
server_log("Finished loading paper models\n")
return paper_models
