def update_model(self, training_data, user_ls):
'''
update parameters for users in user_ls
'''
filtered_data = {u:training_data[u] for u in user_ls}
new_param = process_latencies(to_lat_dict(filtered_data),
lambda x: stats.gamma.fit(
x, floc=0),
lambda: (-1., -1., -1.)
)
for u in user_ls:
self.params[u] = new_param[u]
def estimate_model(self, training_data, val_data):
self.params = process_latencies(to_lat_dict(training_data),
lambda x: stats.gamma.fit(
x, floc=0),
lambda: (-1.,-1.,-1.)
)
def estimate_model(self, training_data, val_data):
self.params = to_lat_dict(training_data)
def update_model(self, training_data, user_ls):
'''
optimization could be made to only merge user_ls's
latency dictionaries.
'''
self.params = to_lat_dict(training_data)
def estimate_model(self,inner_train,inner_val):
self.inn_train_model = process_latencies(
to_lat_dict(inner_train),kdensity,lambda: 0)
self.inn_val_model = process_latencies(to_lat_dict(inner_val),kdensity,lambda: 0)