drzewo decyzyjne i implementacja klasyfikatora

src/agent.py

@@ -61,7 +61,7 @@ class HousePOI:
         self.truck_fullness = Interface.truck_fullness.fullness
         self.payment = payment
         self.bin_fullness = bin_fullness
-        self.trash_type = guessed_trash_type
+        self.trash_type = guessed_trash_type[0]
 
         if self.season == 'autumn':
             season_autumn = 1
@@ -114,11 +114,11 @@ class HousePOI:
         if outcome == 1:
             Interface.truck_fullness.fullness[trash_type] += 0.25  # 0.25
 
-        debug(dec_tree, outcome, trash_type, guessed_trash_type, self.season, self.day, self.bin_fullness,
+        debug(dec_tree, outcome, trash_type, guessed_trash_type[0], guessed_trash_type[1], self.season, self.day, self.bin_fullness,
               self.truck_fullness)
 
 
-def debug(dec_tree, outcome, trash_type, guessed_trash_type, season, day, bin_fullness, truck_fullness):
+def debug(dec_tree, outcome, trash_type, guessed_trash_type, filename, season, day, bin_fullness, truck_fullness):
     HOUSE_NUMBER['House_number'] += 1
     print("Domek nr: " + str(HOUSE_NUMBER['House_number']))
     if outcome == 1:
@@ -128,6 +128,7 @@ def debug(dec_tree, outcome, trash_type, guessed_trash_type, season, day, bin_fu
     print("Pora: " + season)
     print("Dzień: " + day)
     print("Typ śmieci: " + trash_type)
+    print("Nazwa pliku: " + trash_type + str(filename))
     print("Zaobserwowane: " + guessed_trash_type)
     print("Zapełnienie domu: " + str(bin_fullness))
     print(truck_fullness)

src/decisiontree.py

@@ -0,0 +1,30 @@
+import pandas as pd
+from sklearn import tree
+from sklearn.tree import DecisionTreeClassifier
+
+
+def encode_category(category, categories):
+    encoded = pd.get_dummies(categories).astype(int)
+    if category in encoded.columns:
+        return encoded[category].values.tolist()
+    else:
+        return [0] * len(encoded.columns)
+
+
+data = pd.read_csv('tree_data.csv')
+data_encoded = pd.get_dummies(data, columns=['trash_types', 'season', 'day'])
+
+X = data_encoded.drop('decision', axis=1)
+y = data_encoded['decision']
+
+clf = DecisionTreeClassifier()
+decision_tree = clf.fit(X, y)
+
+with open('./tree_in_txt.txt', "w") as file:
+    file.write(tree.export_text(decision_tree, feature_names=X.columns.tolist()))
+
+
+def decision(conditions):
+    new_conditions = pd.DataFrame(conditions)
+    new_decision = decision_tree.predict(new_conditions)
+    return new_decision

src/trashtype.py

@@ -0,0 +1,48 @@
+import tensorflow as tf
+import numpy as np
+import matplotlib.pyplot as plt
+import os
+import PIL
+import random
+from PIL import Image
+
+
+def trash_type_def(trash_type):
+    img_num = 0
+    img_height, img_width = 180, 180
+
+    if trash_type == 'biological':
+        img_num = random.randint(1, 985)
+    if trash_type == 'glass':
+        img_num = random.randint(1, 501)
+    if trash_type == 'paper':
+        img_num = random.randint(1, 594)
+    if trash_type == 'plastic':
+        img_num = random.randint(1, 482)
+
+    filename = trash_type + str(img_num)
+    class_names = ['biological', 'glass', 'paper', 'plastic']
+    trash_classification_model = tf.keras.models.load_model('../trash-classification/m_with_data_augmentation.h5')
+    test_image_path = "../trash-classification/trash/test/"+trash_type+"/"+filename+".jpg"
+
+    img = tf.keras.utils.load_img(
+        test_image_path, target_size=(img_height, img_width)
+    )
+    img_array = tf.keras.utils.img_to_array(img)
+    img_array = tf.expand_dims(img_array, 0)
+
+    predictions = trash_classification_model.predict(img_array)
+    score = tf.nn.softmax(predictions[0])
+
+    name = class_names[np.argmax(score)]
+
+    file_path = "../trash-classification/trash/test/"+trash_type+"/"+filename+".jpg"
+    img_show = Image.open(file_path)
+    img_numpy_format = np.asarray(img_show)
+    plt.imshow(img_numpy_format)
+    plt.suptitle(filename)
+    plt.draw()
+    plt.pause(1)
+    plt.close()
+
+    return [name, img_num]

src/tree.py

@@ -0,0 +1,67 @@
+import csv
+import random
+import pandas as pd
+
+truck_fullness = [0, 1]
+trash_types = ['paper', 'plastic', 'glass', 'mixed']
+payment = [0, 1]
+bin_fullness = [0, 1]
+day = ['monday', 'wednesday', 'friday']
+season = ['spring', 'summer', 'autumn', 'winter']
+dump_fullness = [0, 1]
+trash = [0, 1]
+# spring: monday = plastic, wednesday = glass, friday = biological & paper
+# summer: monday = glass, wednesday = biological & plastic, friday = paper
+# autumn: monday = biological & plastic, wednesday = glass, friday = paper
+# winter: monday = paper, wednesday = biological & plastic, friday = glass
+
+with open('tree_data.csv', 'w', newline='') as csvfile:
+    writer = csv.writer(csvfile)
+
+    writer.writerow(
+        ['dump_fullness', 'truck_fullness', 'trash', 'payment', 'bin_fullness', 'trash_types', 'season', 'day',
+         'decision'])
+    rows = set()
+
+    positives = [(0, 0, 1, 1, 1, 'glass', 'spring', 'wednesday', 1),
+                 (0, 0, 1, 1, 1, 'glass', 'summer', 'monday', 1),
+                 (0, 0, 1, 1, 1, 'glass', 'autumn', 'wednesday', 1),
+                 (0, 0, 1, 1, 1, 'glass', 'winter', 'friday', 1),
+
+                 (0, 0, 1, 1, 1, 'paper', 'spring', 'friday', 1),
+                 (0, 0, 1, 1, 1, 'paper', 'summer', 'friday', 1),
+                 (0, 0, 1, 1, 1, 'paper', 'autumn', 'friday', 1),
+                 (0, 0, 1, 1, 1, 'paper', 'winter', 'monday', 1),
+
+                 (0, 0, 1, 1, 1, 'plastic', 'spring', 'monday', 1),
+                 (0, 0, 1, 1, 1, 'plastic', 'summer', 'wednesday', 1),
+                 (0, 0, 1, 1, 1, 'plastic', 'autumn', 'monday', 1),
+                 (0, 0, 1, 1, 1, 'plastic', 'winter', 'wednesday', 1),
+
+                 (0, 0, 1, 1, 1, 'mixed', 'spring', 'friday', 1),
+                 (0, 0, 1, 1, 1, 'mixed', 'summer', 'wednesday', 1),
+                 (0, 0, 1, 1, 1, 'mixed', 'autumn', 'monday', 1),
+                 (0, 0, 1, 1, 1, 'mixed', 'winter', 'wednesday', 1)]
+
+    for positive in positives:
+        rows.add(positive)
+
+    while len(rows) < 200:
+        rand_1 = random.choice(dump_fullness)
+        rand_2 = random.choice(truck_fullness)
+        rand_3 = random.choice(trash)
+        rand_4 = random.choice(payment)
+        rand_5 = random.choice(bin_fullness)
+        rand_6 = random.choice(trash_types)
+        rand_7 = random.choice(season)
+        rand_8 = random.choice(day)
+
+        rows.add((rand_1, rand_2, rand_3, rand_4, rand_5, rand_6, rand_7, rand_8, 0))
+
+    rows_tuple = tuple(rows)
+
+    for elem in rows_tuple:
+        writer.writerow(elem)
+
+dt = pd.read_csv('tree_data.csv')
+print(dt.head())