Merge pull request #29 from LuposX/development

Development

PYPI/setup.py

@@ -22,7 +22,6 @@
     long_description_content_type="text/markdown",
 
 	# What is in our module/scripts
-	# py_modules=["misc_libary", "polynomial_regression_libary", "linear_regression_libary"],
 	scripts=["boston_housing_prediction/__main__.py"],
 	packages=["boston_housing_prediction"],
     entry_points = {

README.md

@@ -31,9 +31,15 @@ To see the help(for extra options) do:
 $ python -m boston_housing_prediction -h
 ```  
 
-<!--_For more examples and usage, please refer to the [Wiki][wiki]._-->
+<!--__For more examples and usage, please refer to the [Wiki][wiki]._-->
 
 ## Release History
+*   0.2.3
+    *   choosing alpha is now possible
+    *   you can leave inputs empty and it will choose default values
+    *   bugfixes like #25 and more
+    *   uploaded pre-trained models that can be downloaded
+
 *   0.2.2
     *   fix negative input for training of model #24
     *   fix negative output of predictions #24

boston_housing_prediction/boston_main.py

@@ -31,34 +31,37 @@ def main():
     # implemented
     parser.add_argument("--v_data", metavar="VISUALIZE_DATA",
                         help="Set it to True if you want to get a visualization of the data.(default: %(default)s)",
-                        type=bool, default=False)
+                        type=str, default="False")
     # implemented
     parser.add_argument("--v_loss", metavar="VISUALIZE_LOSS",
                         help="Set it to True if you want to get a visualization of the loss.(default: %(default)s)",
-                        type=bool, default=False)
+                        type=str, default="False")
     # implemented
     parser.add_argument("--v_model", metavar="VISUALIZE_MODEL",
                         help="Set it to True if you want to get a visualization of the model.(default: %(default)s)",
-                        type=bool, default=False)
+                        type=str, default="False")
     parser.add_argument("--fd", metavar="FEEDBACK",
                         help="Set how much feedback you want.(Choices: %(choices)s)",
                         type=str, choices=["full", "intermediate", "weak", "debug"], default="immediate")
     # implemented
     parser.add_argument("--save", metavar="SAVE_MODEL",
                         help="Set it to True if you want to save the model after training.(default: %(default)s)",
-                        type=bool, default=False)
+                        type=str, default="False")
 
     parser.add_argument("--h_features", metavar="HELP_FEATURES",
                         help="Set it to True if you want to print out the meaning of the features in the dataset.(default: %(default)s)",
-                        type=bool, default=False)
+                        type=str, default="False")
 
     parser.add_argument("--predict_on",
                         help="Set it to False if you dont want to predict after training.(default: %(default)s)",
-                        action='store_false')
+                        type=str, default="True")
 
     # parse the arguments
     args = parser.parse_args()
 
+    # convert our arguments from strings into booleans
+    parse_bool_args(args)
+
     # check if the dataset exist
     if not is_non_zero_file():
         download_dataset()

boston_housing_prediction/linear_regression_libary.py

@@ -10,8 +10,8 @@ def __init__(self, df, args):
         self.bias = 1
 
         # how man epoch we train
-        self.epochs = 30
-        self.alpha = 0.3
+        self.epochs = 40
+        self.alpha = 0.03
         self.train_loss_history = []
         self.test_loss_history = []
         self.x_train_loose = []
@@ -29,17 +29,44 @@ def __init__(self, df, args):
     # training our model
     def train(self) -> None:
 
+        # getting the learning rate for the model
         while True:
             try:
-                # get input for our model
-                epochs = input("Please type the numbers of epoch you want to train: ")
+                # getting the learning rate
+                alpha = input("Please type the value of learning rate you want to use: ") or self.alpha
+
+                # if default epochs value print the value
+                if alpha == self.alpha:
+                    print(str(self.alpha))
+
+                alpha = float(alpha)
+                if 0 < alpha < 1:
+                    self.alpha = alpha
+                    break
                 print(" ")
+                print("Please input a number between 0 and 1 :)")
+            except ValueError:
+                print(" ")
+                print("Invalid Input!")
+
+        # get epochs for our model
+        while True:
+            try:
+                # get epochs for our model
+                epochs = input("Please type the numbers of epoch you want to train: ") or self.epochs
+
+                # if default epochs value print the value
+                if epochs == self.epochs:
+                    print(str(self.epochs))
+
                 epochs = int(epochs)
                 if epochs > 0:
                     self.epochs = epochs
                     break
+                print(" ")
                 print("Please don't input negative numbers :)")
             except ValueError:
+                print(" ")
                 print("Invalid Input!")
 
         start_time = time.time()
@@ -152,7 +179,8 @@ def predic(self, visualize_process, args_normalization) -> None:
                     sys.exit(0)  # exit the script sucessful
                     break
                 else:
-                    rm_input = round(float(rm_input), 20)
+
+                    rm_input = round(float(rm_input or 5), 20)   # or 5 is the default value
 
                     # checks that no negative numbers get entered
                     if rm_input < 0:
@@ -170,11 +198,13 @@ def predic(self, visualize_process, args_normalization) -> None:
                     # check if predicted output is negative
                     if denorm_pred_target < 0:
                         print("-----------------------------------------------------------------------------")
+                        print("Your input of RM:", str(rm_input))
                         print("Warning: the input values doesn't correspond to a real house.")
                         print("-----------------------------------------------------------------------------")
                         print(" ")
                     else:
                         print("-----------------------------------------------------------------------------")
+                        print("Your input of RM:", str(rm_input))
                         print("Is worth about: " + str(denorm_pred_target) + " in 10,000$(GER 10.000$).")
                         print("-----------------------------------------------------------------------------")
                         print(" ")

boston_housing_prediction/misc_libary.py

@@ -206,7 +206,7 @@ def visualize(args, df_data, parameter_list: list) -> None:
     # get points for line
     X = []
     Y = []
-    for i in range(-10, 14):
+    for i in range(5, 20):
         X.append(i * 0.1)
         Y.append(weights_bias[0] * (i * 0.1) + weights_bias[1])
 
@@ -288,3 +288,36 @@ def v_model_poly(x_axis, y_axis, weights_bias, data_train, target_train):
 
     # hide the grid
     ax.grid(False)
+
+
+# convert our arguments from strings into booleans
+def parse_bool_args(args):
+    if args.predict_on == "False" or args.predict_on == "false" or args.predict_on == "false ":
+        args.predict_on = False
+    if args.predict_on == "True" or args.predict_on == "true" or args.predict_on == "true ":
+        args.predict_on = True
+
+    if args.h_features == "False" or args.h_features == "false" or args.h_features == "false ":
+        args.h_features = False
+    if args.h_features == "True" or args.h_features == "true" or args.h_features == "true ":
+        args.h_features = True
+
+    if args.save == "False" or args.save == "false" or args.save == "false ":
+        args.save = False
+    if args.save == "True" or args.save == "true" or args.save == "true ":
+        args.save = True
+
+    if args.v_model == "False" or args.v_model == "false" or args.v_model == "false ":
+        args.v_model = False
+    if args.v_model == "True" or args.v_model == "true" or args.v_model == "true ":
+        args.v_model = True
+
+    if args.v_loss == "False" or args.v_loss == "false" or args.v_loss == "false ":
+        args.v_loss = False
+    if args.v_loss == "True" or args.v_loss == "true" or args.v_loss == "true ":
+        args.v_loss = True
+
+    if args.v_data == "False" or args.v_data == "false" or args.v_data == "false ":
+        args.v_data = False
+    if args.v_data == "True" or args.v_data == "true" or args.v_data == "true ":
+        args.v_data = True

boston_housing_prediction/polynomial_regression_libary.py

@@ -34,18 +34,44 @@ def hypothesis(self, weights, f1, f2, f3, bias):
 
     # training our model
     def train(self) -> None:
+        # getting the learning rate for the model
+        while True:
+            try:
+                # getting the learning rate
+                alpha = input("Please type the value of learning rate you want to use: ") or self.alpha
+
+                # if default epochs value print the value
+                if alpha == self.alpha:
+                    print(str(self.alpha))
+
+                alpha = float(alpha)
+                if 0 < alpha < 1:
+                    self.alpha = alpha
+                    break
+                print(" ")
+                print("Please input a number between 0 and 1 :)")
+            except ValueError:
+                print(" ")
+                print("Invalid Input!")
+
         # exits while loop when right inputs got inserted
         while True:
             try:
                 # get input for our model
-                epochs = input("Please type the numbers of epoch you want to train: ")
-                print(" ")
+                epochs = input("Please type the numbers of epoch you want to train: ") or self.epochs
+
+                # if default epochs value print the value
+                if epochs == self.epochs:
+                    print(str(self.epochs))
+
                 epochs = int(epochs)
                 if epochs > 0:
                     self.epochs = epochs
                     break
+                print(" ")
                 print("Please don't input negative numbers :)")
             except ValueError:
+                print(" ")
                 print("Invalid Input!")
 
         start_time = time.time()   # start timer. To later calculate time needed to train the model
@@ -173,10 +199,11 @@ def predic(self, visualize_process, args_normalization) -> None:
                 print("Type the Values in the following order: ")
                 print("1.RM 2.LSTAT 3.PTRATIO")
                 input_list = []
+                default_values = [6.24,  12.94, 18.52]  # those are the default values when field is left empty. default values corrospond to mean values of feature
                 for i in range(0,3,1):
                     # exits while loop when right inputs got inserted
                     while True:
-                        input_var = input()
+                        input_var = input() or default_values[i]
 
                         if input_var == "quit" or input_var == "Quit":
                             if visualize_process.is_alive():
@@ -210,7 +237,7 @@ def predic(self, visualize_process, args_normalization) -> None:
                 print(" ")
 
                 # typecasting our inputs and rounding them
-                rm_input = round(float(input_list[0]), 4)
+                rm_input = round(float(input_list[0]), 4)  # or 5 is the default value
                 lstat_input = round(float(input_list[1]), 4)
                 ptratio_input = round(float(input_list[2]), 4)
 
@@ -225,21 +252,24 @@ def predic(self, visualize_process, args_normalization) -> None:
                 # denormalization of output
                 denorm_pred_target = round((self.pred_target * df_range[3]) + df_mean[3], 6)
 
-                print(" ")
-                print("The model predicted that a house with the values: ")
-                print("RM :" + str(rm_input))
-                print("LSTAT :" + str(lstat_input))
-                print("PTRATIO :" + str(ptratio_input))
-                print(" ")
-
                 # check if predicted output is negative
                 if denorm_pred_target < 0:
+                    print(" ")
                     print("-----------------------------------------------------------------------------")
+                    print("The model predicted that a house with the values: ")
+                    print("RM :" + str(rm_input))
+                    print("LSTAT :" + str(lstat_input))
+                    print("PTRATIO :" + str(ptratio_input))
                     print("Warning: the input values doesn't correspond to a real house.")
                     print("-----------------------------------------------------------------------------")
                     print(" ")
                 else:
+                    print(" ")
                     print("-----------------------------------------------------------------------------")
+                    print("The model predicted that a house with the values: ")
+                    print("RM :" + str(rm_input))
+                    print("LSTAT :" + str(lstat_input))
+                    print("PTRATIO :" + str(ptratio_input))
                     print("Is worth about: " + str(denorm_pred_target) + " in 10,000$(GER 10.000$).")
                     print("-----------------------------------------------------------------------------")
                     print(" ")

pre_trained_models/linear_regression_housing_weights.csv

@@ -0,0 +1,3 @@
+weight_bias
+0.9122710219927018
+-0.7175949851060076

pre_trained_models/polynomial_regression_housing_weights.txt

@@ -0,0 +1,10 @@
+0.3577673944247428
+0.7155347888494856
+0.3385774036005275
+-0.3423671694707663
+-0.6847343389415326
+0.3385774036005275
+-0.36920267917045946
+-0.7384053583409189
+0.3719308583089693
+0.05344934752623586

testing_linear.txt

@@ -1,5 +1,7 @@
-50
+0.03
+100
+4.5
 7
 3
--5
+20
 quit

testing_polynomial.txt

@@ -1,6 +1,6 @@
 0.03
 100
+4.5
 7
 3
-20
 quit