Merge branch 'gluonts-prophet' into 'morphemic-rc2.0'

Fixing the FCR prophet error and code refactoring

See merge request !253

forecasting_gluonts/gluonts_forecaster.py

@@ -5,11 +5,8 @@ import itertools
 from sklearn import preprocessing
 from math import log
 from math import exp
-
-pd.set_option('display.max_row', 500)
 import itertools
 from sklearn.model_selection import ParameterGrid
-# from dataset_maker import CSVData
 from time import time
 from time import sleep
 from datetime import datetime
@@ -36,19 +33,20 @@ from sklearn.model_selection import ParameterGrid
 import statistics
 import math
 
+pd.set_option('display.max_row', 500)
 directory_path = "/morphemic_project/"
-pd.options.mode.chained_assignment = None
+pd.options.mode.chained_assignment = None 
 
 
-def data_preprocessing(dataset,metric):
+def data_preprocessing(dataset, metric):
     gluonts_dataset = pd.DataFrame(columns=['ds', 'y'])
     gluonts_dataset['y'] = dataset[metric]
     gluonts_dataset['ds'] = dataset['ems_time']
 
     i = 0
-    while (i<len(gluonts_dataset)):
+    while i<len(gluonts_dataset):
 
-        if (gluonts_dataset['y'][i] == "None"):
+        if gluonts_dataset['y'][i] == "None":
             gluonts_dataset['y'][i] = np.nan
         i += 1
 
@@ -71,24 +69,17 @@ def train(metric):
     dataset = pd.read_csv(data_file_path)
     while len(dataset) < 50:
         dataset = pd.read_csv(data_file_path)
-        logging.debug("Waiting for dataset to be loaded")
+        logging.debug(f"Waiting for dataset to be loaded")
         sleep(30)
-
     gluonts_dataset = data_preprocessing(dataset,metric)
-
     size = len(gluonts_dataset)
-
-
-    logging.debug("STARTED TRAINING FOR: " + metric)
+    logging.debug(f"STARTED TRAINING FOR: {metric}")
 
     # splitting to train and test
     test_percentage = 0.2
     training_window_size = int(len(gluonts_dataset) - (len(gluonts_dataset) * test_percentage))
-
-    train = gluonts_dataset[:training_window_size]
-    validation = gluonts_dataset[training_window_size:]
-
-    train_time = train['ds'][training_window_size - 1]
+    training_dataset = gluonts_dataset[:training_window_size]
+    train_time = training_dataset['ds'][training_window_size - 1]
     validation_time = gluonts_dataset['ds'][size - 1]
     freq = '1min'
 
@@ -142,7 +133,7 @@ def train(metric):
         )
         predictor = estimator.train(training_data=train_ds)
         forecast_it, ts_it = make_evaluation_predictions(
-            dataset=validation_ds,  # validationdataset
+            dataset=validation_ds,  # validation dataset
             predictor=predictor,  # predictor
             num_samples=20,  # number of sample paths we want for evaluation
         )
@@ -178,7 +169,7 @@ def train(metric):
 
     # checking if probabilities file exist
     prob = 0.8
-    if (os.path.isfile(directory_path + 'prob_file.npy')):
+    if os.path.isfile(directory_path + 'prob_file.npy'):
         probs = np.load(directory_path + 'prob_file.npy', allow_pickle='TRUE').item()
         probs[metric] = prob
     else:
@@ -200,8 +191,8 @@ def predict(model, number_of_forward_predictions, prediction_horizon, epoch_star
 
     future = list()
     for i in range(1, number_of_forward_predictions + 1):
-        dateInSec = epoch_start + i * prediction_horizon * 60
-        date = datetime.fromtimestamp(dateInSec)
+        date_in_sec = epoch_start + i * prediction_horizon * 60
+        date = datetime.fromtimestamp(date_in_sec)
         future.append(date)
     future = pd.DataFrame(future)
     future.columns = ['ds']
@@ -224,18 +215,14 @@ def predict(model, number_of_forward_predictions, prediction_horizon, epoch_star
     )
 
     forecasts = list(forecast_it)
-
     forecast_entry = forecasts[0]
-
     logging.debug(forecast_entry)
-
     predictions = forecast_entry.samples
 
     mins = list()
     maxs = list()
     values = list()
 
-    returnDict = dict()
     for i in range(0, number_of_forward_predictions):
         mylist = list()
         for line in predictions:
@@ -248,7 +235,5 @@ def predict(model, number_of_forward_predictions, prediction_horizon, epoch_star
         mins.append(mini)
         maxs.append(maxi)
         values.append(value)
-
-    returnDict = {'mins': mins, 'maxs': maxs, 'values': values}
-
-    return returnDict
+    return_dict = {'mins': mins, 'maxs': maxs, 'values': values}
+    return return_dict

forecasting_gluonts/gluonts_listener.py

@@ -6,8 +6,6 @@ import logging
 import signal
 import threading
 import numpy as np
-
-
 # Libraries required for training and prediction
 import os
 import json
@@ -25,8 +23,6 @@ ACTIVEMQ_PORT = os.environ.get("ACTIVEMQ_PORT")
 
 predictionTimes = dict()
 models = dict()
-# flags = {'avgResponseTime':0 , 'memory': 0}
-
 stop_thread = dict()
 metrics = set()
 metrics_threads = set()
@@ -35,50 +31,38 @@ directory_path = "/morphemic_project/"
 
 def worker(self, body, metric):
     stop_thread[metric] = False
-
-    timestamp = body['timestamp']
     prediction_horizon = body["prediction_horizon"]
     number_of_forward_predictions = body["number_of_forward_predictions"]
     epoch_start = body["epoch_start"]
     predictionTimes[metric] = epoch_start
-
-    while (not os.path.isfile(directory_path + 'models/gluonts_' + metric + ".pkl")):
+    while not os.path.isfile(directory_path + 'models/gluonts_' + metric + ".pkl"):
         sleep(30)
-        logging.debug("Waiting for the trained model for metric: " + metric)
-
-    while (True):
-        if (stop_thread[metric] == True):
+        logging.debug(f"Waiting for the trained model for metric: {metric}")
+    while True:
+        if stop_thread[metric] == True:
             break
         if os.path.isfile(directory_path + 'models/gluonts_' + metric + ".pkl"):
-            logging.debug("Loading the trained model for metric: " + metric)
+            logging.debug(f"Loading the trained model for metric: {metric}")
             with open(directory_path + "models/gluonts_" + metric + ".pkl", 'rb') as f:
                 models[metric] = pickle.load(f)
-
             timestamp = int(time())
-            if (timestamp >= predictionTimes[metric]):
-                logging.debug("Start the prediction for metric: " + metric)
+            if timestamp >= predictionTimes[metric]:
+                logging.debug(f"Start the prediction for metric: {metric}")
                 predictions = gluonts_forecaster.predict(models[metric], number_of_forward_predictions,
                                                          prediction_horizon, epoch_start, metric)
-                # logging.debug(predictions)
                 yhats = predictions['values']
                 yhat_lowers = predictions['mins']
                 yhat_uppers = predictions['maxs']
-
                 prediction_time = epoch_start + prediction_horizon
                 timestamp = int(time())
-
                 # read probabilities file
                 probs = np.load(directory_path + 'prob_file.npy', allow_pickle='TRUE').item()
-
-                logging.debug("Sending predictions for metric: " + metric)
-
+                logging.debug(f"Sending predictions for metric: {metric}")
                 for k in range(0, len(predictions['values'])):
                     yhat = yhats[k]
                     yhat_lower = yhat_lowers[k]
                     yhat_upper = yhat_uppers[k]
-
                     self.connector.send_to_topic('intermediate_prediction.gluonmachines.' + metric,
-
                                                  {
                                                      "metricValue": float(yhat),
                                                      "level": 3,
@@ -91,7 +75,6 @@ def worker(self, body, metric):
                                                      "cloud": "todo",
                                                      "provider": "todo"
                                                  })
-
                     prediction_time = prediction_time + prediction_horizon
                 epoch_start = epoch_start + prediction_horizon
                 sleep(prediction_horizon - 5)
@@ -118,9 +101,8 @@ class Gluonts(morphemic.handler.ModelHandler, messaging.listener.MorphemicListen
         pass
 
     def on_start_forecasting_gluonmachines(self, body):
-        logging.debug("Gluonts Start Forecasting the following metrics :")
         sent_metrics = body["metrics"]
-        logging.debug(sent_metrics)
+        logging.debug(f"Gluonts Start Forecasting the following metrics: {sent_metrics}")
         for metric in sent_metrics:
             if metric not in metrics:
                 metrics.add(metric)
@@ -129,59 +111,46 @@ class Gluonts(morphemic.handler.ModelHandler, messaging.listener.MorphemicListen
                 x = threading.Thread(target=worker, args=(self, body, metric,))
                 x.start()
 
-
-
     def on_metrics_to_predict(self, body):
         dataset_preprocessor = CSVData(APP_NAME)
         data_file_path = os.path.join(os.environ.get("DATA_PATH", "./"), f'{os.environ.get("APP_NAME", "demo")}.csv')
         dataset_preprocessor.prepare_csv()
 
-        while (not os.path.isfile(data_file_path)):
-            logging.debug("Waiting for dataset to be loaded")
+        while not os.path.isfile(data_file_path):
+            logging.debug(f"Waiting for dataset to be loaded")
             sleep(30)
             dataset_preprocessor.prepare_csv()
-
         logging.debug("DATASET DOWNLOADED")
-
         for r in body:
             metric = r['metric']
             if not os.path.isfile(directory_path + 'models/gluonts_' + metric + ".pkl"):
-                logging.debug("Training a GluonTS model for metric : " + metric)
+                logging.debug(f"Training a GluonTS model for metric: {metric}")
                 model = gluonts_forecaster.train(metric)
                 pkl_path = directory_path + "models/gluonts_" + metric + ".pkl"
                 with open(pkl_path, "wb") as f:
                     pickle.dump(model, f)
-                    logging.debug("Model Dumped")
+                    logging.debug(f"Model Dumped")
             metrics.add(metric)
-
         self.connector.send_to_topic("training_models",
                                      {
-
                                          "metrics": list(metrics),
-
                                          "forecasting_method": "gluonmachines",
-
                                          "timestamp": int(time())
-                                     }
-                                     )
-
+                                     })
 
     def on_stop_forecasting_gluonmachines(self, body):
-        logging.debug("Gluonts Stop Forecasting the following metrics :")
-        logging.debug(body["metrics"])
+        logging.debug(f"Gluonts Stop Forecasting the following metrics: {body['metrics']}")
         for metric in body["metrics"]:
             if metric in metrics:
                 metrics_threads.remove(metric)
                 stop_thread[metric] = True
 
-
     def start(self):
-        logging.debug("Staring Gluonts Forecaster")
+        logging.debug(f"Staring Gluonts Forecaster")
         self.run()
         self._run = True
 
-
     def on_disconnected(self):
-        print('Disconnected from ActiveMQ')
+        logging.debug(f'Disconnected from ActiveMQ')
         self.reconnect()
 

forecasting_gluonts/main.py

@@ -4,12 +4,7 @@ import time
 
 logger = logging.getLogger()
 logger.setLevel(logging.DEBUG)
-
 e = gluonts_listener.Gluonts()
-#try:
-#    e.start()
-#except KeyboardInterrupt:
-#    e.stop()
 e.start()
 while True:
 	time.sleep(60)

forecasting_prophet/main.py

@@ -5,15 +5,12 @@ import time
 
 def main():
 	e = prophet_listener.Prophet()
-	#try:
-	#    e.start()
-	#except KeyboardInterrupt:
-	#    e.stop()
 	e.start()
 	while True:
 		time.sleep(60)
 		pass
 
+
 if __name__ == '__main__':
 	logging.config.fileConfig('/morphemic_project/logging.ini', disable_existing_loggers=False)
 	main()

forecasting_prophet/prophet_forecaster.py

@@ -12,8 +12,6 @@ from scipy.stats import boxcox
 from scipy.special import inv_boxcox
 from math import log
 from math import exp
-
-pd.set_option('display.max_row', 500)
 import itertools
 from sklearn.model_selection import ParameterGrid
 from time import time
@@ -24,31 +22,31 @@ import json
 import os
 import math
 import logging.config
-from time import sleep
+from time import sleep 
 
+pd.set_option('display.max_row', 500)
 directory_path = "/morphemic_project/"
 pd.options.mode.chained_assignment = None
 
 # logging.config.fileConfig('logging.ini', disable_existing_loggers=False)
 
+
 def train(metric):
     data_file_path = os.path.join(os.environ.get("DATA_PATH", "./"), f'{os.environ.get("APP_NAME", "demo")}.csv')
     init = 0
     h = 0
     p = 0
-    while (init <= 0 or h <= 0 or p <= 0):
+    while init <= 0 or h <= 0 or p <= 0:
         dataset = pd.read_csv(data_file_path)
-        # hyperparameter tuning and cross validation
-        # should be generic
         for i in range(0, len(dataset['ems_time'])):
             dataset['ems_time'][i] = datetime.fromtimestamp(dataset['ems_time'][i])
 
         t1 = int(dataset['ems_time'][0].strftime('%s'))
         t2 = int(dataset['ems_time'][len(dataset) - 1].strftime('%s'))
 
-        timeDiffInSec = int(t2 - t1)
-        timeDiffInMin = timeDiffInSec / 60
-        init = int(timeDiffInMin / 3)
+        time_diff_in_sec = int(t2 - t1)
+        time_diff_in_min = time_diff_in_sec / 60
+        init = int(time_diff_in_min / 3)
         h = int(init / 3)
         p = int(h / 2)
 
@@ -56,21 +54,22 @@ def train(metric):
         horizon = str(h) + " minutes"
         period = str(p) + " minutes"
 
-        logging.debug("initial : ", initial)
-        logging.debug("horizon : ", horizon)
-        logging.debug("period : ", period)
-        if (init <= 0 or h <= 0 or p <= 0):
-            logging.debug("Dataset is not enough for training the model")
+        logging.debug(f"period: {period}")
+        logging.debug(f"horizon: {horizon}")
+        logging.debug(f"initial: {initial}")
+
+        if init <= 0 or h <= 0 or p <= 0:
+            logging.debug(f"Dataset is not enough for training the model")
             sleep(30)
 
-    #Data preprocessing
+    # Data preprocessing
     prophet_dataset = pd.DataFrame(columns=['ds', 'y'])
     prophet_dataset['y'] = dataset[metric]
     prophet_dataset['ds'] = dataset['ems_time']
 
     i = 0
-    while (i<len(prophet_dataset)):
-        if (prophet_dataset['y'][i] == "None"):
+    while i < len(prophet_dataset):
+        if prophet_dataset['y'][i] == "None":
             prophet_dataset['y'][i] = np.nan
         i += 1
 
@@ -78,11 +77,11 @@ def train(metric):
     prophet_dataset = prophet_dataset.ffill(axis ='rows')
     prophet_dataset = prophet_dataset.fillna(0)
 
-    logging.debug("Cleaned prophet_dataset : ", prophet_dataset)
+    logging.debug(f"Cleaned prophet_dataset: {prophet_dataset}")
 
-    #Model trainig
-    logging.debug("STARTED TRAINING FOR: " + metric)
-    train = prophet_dataset[:len(prophet_dataset)]
+    # Model training
+    logging.debug(f"STARTED TRAINING FOR: {metric}")
+    training_dataset = prophet_dataset[:len(prophet_dataset)]
 
     '''
     changepoint_prior_scale  = [0.1,0.2,0.3,0.4,0.5]
@@ -109,7 +108,7 @@ def train(metric):
     grid = ParameterGrid(param_grid)
     cnt = 0
     for p in grid:
-        cnt = cnt + 1
+        cnt += 1
 
     all_params = [dict(zip(param_grid.keys(), v)) for v in itertools.product(*param_grid.values())]
     rmses = []  # Store the RMSEs for each params here
@@ -119,7 +118,7 @@ def train(metric):
 
     # use cross validation to evaluate all parameters
     for params in all_params:
-        m = Prophet(**params).fit(train)  # Fit model with given params
+        m = Prophet(**params).fit(training_dataset)  # Fit model with given params
         df_cv = cross_validation(m, initial=initial, period=period, horizon=horizon)
         df_p = performance_metrics(df_cv, rolling_window=1)
         rmses.append(df_p['rmse'].values[0])
@@ -159,9 +158,9 @@ def train(metric):
         # holidays_prior_scale = parameters['holidays_prior_scale'][0],
         # mcmc_samples = parameters['mcmc_samples'][0]
     )
-    final_model.fit(train)
+    final_model.fit(training_dataset)
     # checking if probabilities file exist
-    if (os.path.isfile(directory_path + 'prob_file.npy')):
+    if os.path.isfile(directory_path + 'prob_file.npy'):
         probs = np.load(directory_path + 'prob_file.npy', allow_pickle='TRUE').item()
         probs[metric] = prob
 
@@ -180,8 +179,8 @@ def train(metric):
 def predict(model, number_of_forward_predictions, prediction_horizon, epoch_start):
     future = list()
     for i in range(1, number_of_forward_predictions + 1):
-        dateInSec = epoch_start + i * prediction_horizon
-        date = datetime.fromtimestamp(dateInSec)
+        date_in_sec = epoch_start + i * prediction_horizon
+        date = datetime.fromtimestamp(date_in_sec)
         future.append(date)
     future = pd.DataFrame(future)
     future.columns = ['ds']

forecasting_prophet/prophet_listener.py

@@ -19,10 +19,6 @@ from time import sleep
 from dataset_maker import CSVData
 from multiprocessing import Process
 
-
-#logging.config.fileConfig('logging.ini', disable_existing_loggers=False)
-
-
 APP_NAME = os.environ.get("APP_NAME")
 ACTIVEMQ_USER = os.environ.get("ACTIVEMQ_USER")
 ACTIVEMQ_PASSWORD = os.environ.get("ACTIVEMQ_PASSWORD")
@@ -31,84 +27,69 @@ ACTIVEMQ_PORT = os.environ.get("ACTIVEMQ_PORT")
 
 predictionTimes = dict()
 models = dict()
-#flags = {'avgResponseTime':0 , 'memory': 0}
-metrics_processes=dict()
+metrics_processes = dict()
 metrics = set()
 directory_path = "/morphemic_project/"
 
-def worker(self,body,metric):
 
+def worker(self, body, metric):
     timestamp = body['timestamp']
     prediction_horizon = body["prediction_horizon"]
     number_of_forward_predictions = body["number_of_forward_predictions"]   
-    epoch_start= body["epoch_start"]
+    epoch_start = body["epoch_start"]
     predictionTimes[metric] = epoch_start
-    messages=list()
-    f=0
-    while (not os.path.isfile(directory_path+'models/prophet_'+metric+".pkl")):
+    messages = list()
+    f = 0
+    while not os.path.isfile(directory_path + 'models/prophet_' + metric + ".pkl"):
         sleep(30)
-        logging.debug("Waiting for the trained model for metric: " + metric)
+        logging.debug(f"Waiting for the trained model for metric: {metric}")
 
-    while(True):
-        #if flags[metric] == 0:
-            #epoch_start = predictionTimes[metric]
-            #flags[metric] = 1
-            #load the model
+    while True:
         with open(directory_path+"models/prophet_"+metric+".pkl", 'rb') as f:
             models[metric] = pickle.load(f)
         timestamp = int(time())   
-        if (timestamp >= predictionTimes[metric]):
-            predictions=prophet_forecaster.predict(models[metric] , number_of_forward_predictions , prediction_horizon , epoch_start)
+        if timestamp >= predictionTimes[metric]:
+            predictions = prophet_forecaster.predict(models[metric], number_of_forward_predictions, prediction_horizon, epoch_start)
             yhats = predictions['yhat'].values.tolist()
             yhat_lowers = predictions['yhat_lower'].values.tolist()
             yhat_uppers = predictions['yhat_upper'].values.tolist()
-            
-            prediction_time= epoch_start+ prediction_horizon
-           # change it to the time of the start_forecasting was sent
-            
-            #read probabilities file
-            probs = np.load(directory_path+'prob_file.npy' , allow_pickle='TRUE').item()
+            prediction_time = epoch_start + prediction_horizon
+            # change it to the time of the start_forecasting was sent
+            # read probabilities file
+            probs = np.load(directory_path+'prob_file.npy', allow_pickle='TRUE').item()
 
             for k in range(0,len(predictions['yhat'].values.tolist())):
                 yhat = yhats[k]
                 yhat_lower = yhat_lowers[k]
                 yhat_upper = yhat_uppers[k]
-                
-                #wait until epoch_start to send
-                
+                # wait until epoch_start to send
                 message = {
                     "metricValue": yhat,
                     "level": 3,
                     "timestamp": timestamp,
                     "probability": probs[metric],
-                    "confidence_interval" : [yhat_lower,yhat_upper],
+                    "confidence_interval": [yhat_lower, yhat_upper],
                     "horizon": prediction_horizon,
-                    "predictionTime" : int(prediction_time), # 
+                    "predictionTime": int(prediction_time), #
                     "refersTo": "todo",
                     "cloud": "todo",
                     "provider": "todo"  
                     }
-                
-                
-                
                 self.connector.send_to_topic('intermediate_prediction.prophet.'+metric, message) 
-                prediction_time=prediction_time + prediction_horizon
-            epoch_start = epoch_start+ prediction_horizon
+                prediction_time = prediction_time + prediction_horizon
+            epoch_start = epoch_start + prediction_horizon
             sleep(prediction_horizon)
 
 
-class Prophet(morphemic.handler.ModelHandler,messaging.listener.MorphemicListener):
+class Prophet(morphemic.handler.ModelHandler, messaging.listener.MorphemicListener):
     id = "prophet"
     metrics = set()
 
     def __init__(self):
-        self._run =  False
+        self._run = False
         self.connector = messaging.morphemic.Connection(ACTIVEMQ_USER,ACTIVEMQ_PASSWORD, host=ACTIVEMQ_HOSTNAME, port=ACTIVEMQ_PORT)
 
     def run(self):
-        #logging.debug("setting up")
-        #sleep(180)
-        #logging.debug("starting the connection to ActiveMQ")
         self.connector.connect()
         self.connector.set_listener(self.id, self)
         self.connector.topic("start_forecasting.prophet", self.id)
@@ -116,71 +97,65 @@ class Prophet(morphemic.handler.ModelHandler,messaging.listener.MorphemicListene
         self.connector.topic("metrics_to_predict", self.id)
 
     def reconnect(self):
-        print('Reconnecting to ActiveMQ')
+        logging.debug(f"Reconnecting to ActiveMQ")
         self.connector.disconnect()
         self.run()
         pass
 
     def on_start_forecasting_prophet(self, body):
-        logging.debug("Prophet Start Forecasting the following metrics :") 
+        logging.debug(f"Prophet Start Forecasting the following metrics:")
         sent_metrics = body["metrics"]
         logging.debug(sent_metrics)
         for metric in sent_metrics:
             if metric not in metrics:
                 metrics.add(metric)
-            #thread = threading.Thread(target=worker , args=(self, body, metric,))
-            if  metric not in metrics_processes:
+            if metric not in metrics_processes:
                 metrics_processes[metric] = Process(target=worker, args=(self, body, metric,))
                 metrics_processes[metric] .start()
 
     def on_metrics_to_predict(self, body):
-        #getting data from datasetmaker
+        # getting data from dataset maker
         dataset_preprocessor = CSVData(APP_NAME)
         data_file_path = os.path.join(os.environ.get("DATA_PATH", "./"), f'{os.environ.get("APP_NAME", "demo")}.csv')
         dataset_preprocessor.prepare_csv()
-
-        while (not os.path.isfile(data_file_path)):