noshot 11.0.0__py3-none-any.whl → 13.0.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (93) hide show
  1. noshot/data/DLE FSD BDA/DLE/DLE 1 (Json)/1. DNN (Image Classification).ipynb +389 -0
  2. noshot/data/DLE FSD BDA/DLE/DLE 1 (Json)/2. DNN vs CNN.ipynb +516 -0
  3. noshot/data/DLE FSD BDA/DLE/DLE 1 (Json)/3. CNN (Object Detecrion).ipynb +259 -0
  4. noshot/data/DLE FSD BDA/DLE/DLE 1 (Json)/4. FCN (Image Segmentaion).ipynb +274 -0
  5. noshot/data/DLE FSD BDA/DLE/DLE 2 (tim stan s)/1.1 DNN (Pytorch).ipynb +164 -0
  6. noshot/data/DLE FSD BDA/DLE/DLE 2 (tim stan s)/1.2 DNN (Tensorflow).ipynb +94 -0
  7. noshot/data/DLE FSD BDA/DLE/DLE 2 (tim stan s)/1.3 DNN (Image Classification).ipynb +134 -0
  8. noshot/data/DLE FSD BDA/DLE/DLE 2 (tim stan s)/2.1 DNN vs CNN.ipynb +127 -0
  9. noshot/data/DLE FSD BDA/DLE/DLE 2 (tim stan s)/2.2 DNN vs CNN.ipynb +123 -0
  10. noshot/data/DLE FSD BDA/DLE/DLE 2 (tim stan s)/4. FCNN (Image Segmentation).ipynb +108 -0
  11. noshot/data/DLE FSD BDA/DLE/DLE 2 (tim stan s)/Lab Excercise (Training DNN).ipynb +646 -0
  12. noshot/data/DLE FSD BDA/DLE/DLE 2 (tim stan s)/Load-Images.ipynb +553 -0
  13. noshot/data/DLE FSD BDA/DLE/DLE 3 (sonic boy)/Ex1.ipynb +216 -0
  14. noshot/data/DLE FSD BDA/DLE/DLE 3 (sonic boy)/Ex2.ipynb +195 -0
  15. noshot/data/DLE FSD BDA/DLE/DLE 3 (sonic boy)/Ex3.ipynb +427 -0
  16. noshot/data/DLE FSD BDA/DLE/DLE 3 (sonic boy)/Ex4.ipynb +186 -0
  17. noshot/data/DLE FSD BDA/DLE/DLE 4 (senior)/Exp01/DNN Ex No 1.ipynb +398 -0
  18. noshot/data/DLE FSD BDA/DLE/DLE 4 (senior)/Exp01/Ex No 1 Build in dataset.ipynb +171 -0
  19. noshot/data/DLE FSD BDA/DLE/DLE 4 (senior)/Exp01/Exp1-Short-DL_ANN_ImageClassification.ipynb +401 -0
  20. noshot/data/DLE FSD BDA/DLE/DLE 4 (senior)/Exp01/OR GATE .ipynb +8511 -0
  21. noshot/data/DLE FSD BDA/DLE/DLE 4 (senior)/Exp02/Exp2-Short-DL_CNN_ImageClassification.ipynb +737 -0
  22. noshot/data/DLE FSD BDA/DLE/DLE 4 (senior)/Exp03/DL-Ex3-RNN.ipynb +591 -0
  23. noshot/data/DLE FSD BDA/DLE/DLE 4 (senior)/Exp04/Ex no 4.ipynb +551 -0
  24. noshot/main.py +3 -3
  25. {noshot-11.0.0.dist-info → noshot-13.0.0.dist-info}/METADATA +1 -1
  26. noshot-13.0.0.dist-info/RECORD +32 -0
  27. noshot/data/ML TS XAI/ML/CNN(Image_for_Folders_5).ipynb +0 -201
  28. noshot/data/ML TS XAI/ML/CNN(Image_form_Folder_2).ipynb +0 -201
  29. noshot/data/ML TS XAI/ML/Json Codes/ML LAB CIA 2.ipynb +0 -409
  30. noshot/data/ML TS XAI/ML/ML 1/1. EDA-PCA (Balance Scale Dataset).ipynb +0 -147
  31. noshot/data/ML TS XAI/ML/ML 1/1. EDA-PCA (Rice Dataset).ipynb +0 -181
  32. noshot/data/ML TS XAI/ML/ML 1/10. HMM Veterbi.ipynb +0 -152
  33. noshot/data/ML TS XAI/ML/ML 1/2. KNN (Balance Scale Dataset).ipynb +0 -117
  34. noshot/data/ML TS XAI/ML/ML 1/2. KNN (Iris Dataset).ipynb +0 -156
  35. noshot/data/ML TS XAI/ML/ML 1/2. KNN (Sobar-72 Dataset).ipynb +0 -215
  36. noshot/data/ML TS XAI/ML/ML 1/3. LDA (Balance Scale Dataset).ipynb +0 -78
  37. noshot/data/ML TS XAI/ML/ML 1/3. LDA (NPHA Doctor Visits Dataset).ipynb +0 -114
  38. noshot/data/ML TS XAI/ML/ML 1/4. Linear Regression (Machine Dataset).ipynb +0 -115
  39. noshot/data/ML TS XAI/ML/ML 1/4. Linear Regression (Real Estate Dataset).ipynb +0 -146
  40. noshot/data/ML TS XAI/ML/ML 1/5. Logistic Regression (Magic04 Dataset).ipynb +0 -130
  41. noshot/data/ML TS XAI/ML/ML 1/5. Logistic Regression (Wine Dataset).ipynb +0 -112
  42. noshot/data/ML TS XAI/ML/ML 1/6. Naive Bayes Classifier (Agaricus Lepiota Dataset).ipynb +0 -118
  43. noshot/data/ML TS XAI/ML/ML 1/6. Naive Bayes Classifier (Wine Dataset).ipynb +0 -89
  44. noshot/data/ML TS XAI/ML/ML 1/7. SVM (Rice Dataset).ipynb +0 -120
  45. noshot/data/ML TS XAI/ML/ML 1/8. FeedForward NN (Sobar72 Dataset).ipynb +0 -262
  46. noshot/data/ML TS XAI/ML/ML 1/9. CNN (Cifar10 Dataset).ipynb +0 -156
  47. noshot/data/ML TS XAI/ML/ML 2/1. PCA.ipynb +0 -162
  48. noshot/data/ML TS XAI/ML/ML 2/10. CNN.ipynb +0 -100
  49. noshot/data/ML TS XAI/ML/ML 2/11. HMM.ipynb +0 -336
  50. noshot/data/ML TS XAI/ML/ML 2/2. KNN.ipynb +0 -149
  51. noshot/data/ML TS XAI/ML/ML 2/3. LDA.ipynb +0 -132
  52. noshot/data/ML TS XAI/ML/ML 2/4. Linear Regression.ipynb +0 -86
  53. noshot/data/ML TS XAI/ML/ML 2/5. Logistic Regression.ipynb +0 -115
  54. noshot/data/ML TS XAI/ML/ML 2/6. Naive Bayes (Titanic).ipynb +0 -196
  55. noshot/data/ML TS XAI/ML/ML 2/6. Naive Bayes (Wine).ipynb +0 -98
  56. noshot/data/ML TS XAI/ML/ML 2/7. SVM Linear.ipynb +0 -109
  57. noshot/data/ML TS XAI/ML/ML 2/8. SVM Non-Linear.ipynb +0 -195
  58. noshot/data/ML TS XAI/ML/ML 2/9. FNN With Regularization.ipynb +0 -189
  59. noshot/data/ML TS XAI/ML/ML 2/9. FNN Without Regularization.ipynb +0 -197
  60. noshot/data/ML TS XAI/ML/ML 2/All in One Lab CIA 1 Q.ipynb +0 -1087
  61. noshot/data/ML TS XAI/ML/ML 3 (Latest)/1. PCA EDA.ipynb +0 -274
  62. noshot/data/ML TS XAI/ML/ML 3 (Latest)/10. CNN.ipynb +0 -170
  63. noshot/data/ML TS XAI/ML/ML 3 (Latest)/11. HMM 2.ipynb +0 -1087
  64. noshot/data/ML TS XAI/ML/ML 3 (Latest)/11. HMM 3.ipynb +0 -178
  65. noshot/data/ML TS XAI/ML/ML 3 (Latest)/11. HMM 4.ipynb +0 -185
  66. noshot/data/ML TS XAI/ML/ML 3 (Latest)/11. HMM.ipynb +0 -106
  67. noshot/data/ML TS XAI/ML/ML 3 (Latest)/2. KNN.ipynb +0 -177
  68. noshot/data/ML TS XAI/ML/ML 3 (Latest)/3. LDA.ipynb +0 -195
  69. noshot/data/ML TS XAI/ML/ML 3 (Latest)/4. Linear Regression.ipynb +0 -267
  70. noshot/data/ML TS XAI/ML/ML 3 (Latest)/5. Logistic Regression.ipynb +0 -104
  71. noshot/data/ML TS XAI/ML/ML 3 (Latest)/6. Bayesian Classifier.ipynb +0 -109
  72. noshot/data/ML TS XAI/ML/ML 3 (Latest)/7. SVM.ipynb +0 -220
  73. noshot/data/ML TS XAI/ML/ML 3 (Latest)/8. MLP.ipynb +0 -99
  74. noshot/data/ML TS XAI/ML/ML 3 (Latest)/9. Ridge - Lasso.ipynb +0 -211
  75. noshot/data/ML TS XAI/ML/ML 3 (Latest)/9. Ridge Lasso 2.ipynb +0 -99
  76. noshot/data/ML TS XAI/ML/ML 3 (Latest)/Image Load Example.ipynb +0 -118
  77. noshot/data/ML TS XAI/ML/ML 3 (Latest)/Updated_Untitled.ipynb +0 -603
  78. noshot/data/ML TS XAI/ML/ML Lab AllinOne.ipynb +0 -961
  79. noshot/data/ML TS XAI/ML/ML Lab H Sec/1. Iris Dataset (Softmax vs Sigmoid).ipynb +0 -231
  80. noshot/data/ML TS XAI/ML/ML Lab H Sec/2. Student Dataset (Overfit vs Regularized).ipynb +0 -269
  81. noshot/data/ML TS XAI/ML/ML Lab H Sec/3. Insurance Target Categorical (Overfit vs Regularized).ipynb +0 -274
  82. noshot/data/ML TS XAI/ML/ML Lab H Sec/3. Insurance Target Numerical (Overfit vs Regularized).ipynb +0 -263
  83. noshot/data/ML TS XAI/ML/ML Lab H Sec/4. Smart House System HMM.ipynb +0 -198
  84. noshot/data/ML TS XAI/ML/ML Lab H Sec/5. Fraud Detection System HMM.ipynb +0 -201
  85. noshot/data/ML TS XAI/ML/ML Lab H Sec/insurance.csv +0 -1339
  86. noshot/data/ML TS XAI/ML/ML Lab H Sec/iris1.data +0 -151
  87. noshot/data/ML TS XAI/ML/ML Lab H Sec/student-mat.csv +0 -396
  88. noshot/data/ML TS XAI/ML/ML Lab H Sec/student-por.csv +0 -650
  89. noshot/data/ML TS XAI/ML/Rolls Royce AllinOne.ipynb +0 -691
  90. noshot-11.0.0.dist-info/RECORD +0 -72
  91. {noshot-11.0.0.dist-info → noshot-13.0.0.dist-info}/WHEEL +0 -0
  92. {noshot-11.0.0.dist-info → noshot-13.0.0.dist-info}/licenses/LICENSE.txt +0 -0
  93. {noshot-11.0.0.dist-info → noshot-13.0.0.dist-info}/top_level.txt +0 -0
noshot/data/ML TS XAI/ML/ML Lab AllinOne.ipynb DELETED
@@ -1,961 +0,0 @@
1
- {
2
- "cells": [
3
- {
4
- "cell_type": "markdown",
5
- "id": "007cdc19-d98f-4533-a27b-2ea68643fd09",
6
- "metadata": {},
7
- "source": [
8
- "#### __Most plots Related to dataset with categorical variable as output are used in KNN exp__\n",
9
- "#### __But it should also be used for relevant other experiments also like Naive Bayes, Logistic Regression__"
10
- ]
11
- },
12
- {
13
- "cell_type": "markdown",
14
- "id": "c73ca8b2-5a11-488d-97ca-160ad0ff4f18",
15
- "metadata": {},
16
- "source": [
17
- "### __***PCA***__"
18
- ]
19
- },
20
- {
21
- "cell_type": "code",
22
- "execution_count": null,
23
- "id": "e4a62599-bb3d-4b3d-8bcd-ec26893e3921",
24
- "metadata": {},
25
- "outputs": [],
26
- "source": [
27
- "import pandas as pd\n",
28
- "import numpy as np\n",
29
- "import seaborn as sns\n",
30
- "import matplotlib.pyplot as plt\n",
31
- "from sklearn.preprocessing import StandardScaler\n",
32
- "from sklearn.decomposition import PCA\n",
33
- "\n",
34
- "df = pd.read_table('data/balance-scale.csv', delimiter=',')\n",
35
- "print(\"Shape:\", df.shape)\n",
36
- "display(df.head())\n",
37
- "\n",
38
- "X = df.drop(columns='class name')\n",
39
- "y = df['class name']\n",
40
- "\n",
41
- "scaled = StandardScaler().fit_transform(X)\n",
42
- "pca = PCA(n_components=2).fit_transform(scaled)\n",
43
- "\n",
44
- "final = pd.DataFrame(pca, columns=['PC1', 'PC2'])\n",
45
- "final['target'] = df['class name']\n",
46
- "final.head()\n",
47
- "\n",
48
- "sns.countplot(df, x='class name', hue='class name')\n",
49
- "plt.show()\n",
50
- "sns.heatmap(X.corr(), cmap='Blues')\n",
51
- "plt.show()\n",
52
- "sns.pairplot(X)\n",
53
- "plt.show()\n",
54
- "sns.scatterplot(final, x='PC1', y='PC2', hue='target')\n",
55
- "plt.show()"
56
- ]
57
- },
58
- {
59
- "cell_type": "markdown",
60
- "id": "9c779383-f645-4ee2-8bd0-4d9ce2f900fb",
61
- "metadata": {},
62
- "source": [
63
- "### __***KNN***__"
64
- ]
65
- },
66
- {
67
- "cell_type": "code",
68
- "execution_count": null,
69
- "id": "4b52d350-cc91-4505-96b5-8fad7f4eb6f7",
70
- "metadata": {},
71
- "outputs": [],
72
- "source": [
73
- "import numpy as np\n",
74
- "import pandas as pd\n",
75
- "import seaborn as sns\n",
76
- "import matplotlib.pyplot as plt\n",
77
- "from sklearn.preprocessing import StandardScaler\n",
78
- "from sklearn.model_selection import train_test_split, cross_val_score\n",
79
- "from sklearn.decomposition import PCA\n",
80
- "from sklearn.neighbors import KNeighborsClassifier\n",
81
- "from sklearn.metrics import confusion_matrix, classification_report\n",
82
- "from sklearn.metrics import accuracy_score, ConfusionMatrixDisplay\n",
83
- "from mlxtend.plotting import plot_decision_regions\n",
84
- "from scipy.cluster.hierarchy import dendrogram, linkage\n",
85
- "from scipy.spatial import Voronoi, voronoi_plot_2d\n",
86
- "\n",
87
- "df = pd.read_csv(\"data/sobar-72.csv\")\n",
88
- "print(\"Shape:\", df.shape)\n",
89
- "display(df.head())\n",
90
- "\n",
91
- "X = df.drop(columns='ca_cervix')\n",
92
- "y = df['ca_cervix']\n",
93
- "\n",
94
- "X_scaled = StandardScaler().fit_transform(X)\n",
95
- "X_pca = PCA(n_components=2).fit_transform(X_scaled)\n",
96
- "X_train, X_test, y_train, y_test = train_test_split(X_pca, y, \n",
97
- " test_size=0.4,\n",
98
- " random_state=4)\n",
99
- "\n",
100
- "knn = KNeighborsClassifier(n_neighbors=15, metric='euclidean') \n",
101
- "#can be ['euclidean', 'manhattan', 'minkowski']\n",
102
- "knn.fit(X_train, y_train)\n",
103
- "y_pred = knn.predict(X_test)\n",
104
- "\n",
105
- "sns.countplot(df, x='ca_cervix', hue='ca_cervix')\n",
106
- "plt.show()\n",
107
- "\n",
108
- "sns.pairplot(df.iloc[:,:4])\n",
109
- "plt.show()\n",
110
- "\n",
111
- "# dont use this mlxtend module may not be installed in lab\n",
112
- "# ========================================================\n",
113
- "plot_decision_regions(X_train, y_train.values, clf=knn, legend=2)\n",
114
- "plt.xlabel('X')\n",
115
- "plt.ylabel('Y')\n",
116
- "plt.title(f'KNN with K=5 using Euclidean Distance')\n",
117
- "plt.show()\n",
118
- "# ========================================================\n",
119
- "print(classification_report(y_test, y_pred))\n",
120
- "print(\"Accuracy\", accuracy_score(y_test, y_pred))\n",
121
- "ConfusionMatrixDisplay(confusion_matrix(y_test, y_pred)).plot()\n",
122
- "plt.show()\n",
123
- "\n",
124
- "errors = [1 - cross_val_score(KNeighborsClassifier(n_neighbors=k), \n",
125
- " X_train, y_train).mean() for k in range(1, 21)]\n",
126
- "plt.plot(range(1, 21), errors, marker='o')\n",
127
- "plt.title(\"Elbow Method for Optimal k\")\n",
128
- "plt.xlabel(\"k\")\n",
129
- "plt.ylabel(\"Error\")\n",
130
- "plt.show()\n",
131
- "\n",
132
- "vor = Voronoi(X_pca)\n",
133
- "voronoi_plot_2d(vor, show_vertices=False)\n",
134
- "plt.scatter(X_pca[:, 0], X_pca[:, 1], c=y)\n",
135
- "plt.show()\n",
136
- "\n",
137
- "# use this instead for decision boundary graph\n",
138
- "x_min, x_max = X_pca[:,0].min() - 1, X_pca[:,0].max() + 1\n",
139
- "y_min, y_max = X_pca[:,1].min() - 1, X_pca[:,1].max() + 1\n",
140
- "xx, yy = np.meshgrid(np.arange(x_min, x_max, 0.02), \n",
141
- " np.arange(y_min, y_max, 0.02))\n",
142
- "\n",
143
- "Z = knn.predict(np.c_[xx.ravel(), yy.ravel()]).reshape(xx.shape)\n",
144
- "\n",
145
- "plt.figure()\n",
146
- "plt.contourf(xx, yy, Z, cmap=plt.cm.coolwarm, alpha=0.8)\n",
147
- "plt.scatter(X_pca[:,0], X_pca[:,1], c=y, cmap=plt.cm.coolwarm, s=20, \n",
148
- " edgecolors='k')\n",
149
- "plt.title('Decision surface')\n",
150
- "plt.show()"
151
- ]
152
- },
153
- {
154
- "cell_type": "markdown",
155
- "id": "b9a6637d-95a0-45c0-8527-0febe5ce29b4",
156
- "metadata": {},
157
- "source": [
158
- "### __***LDA***__"
159
- ]
160
- },
161
- {
162
- "cell_type": "code",
163
- "execution_count": null,
164
- "id": "6fcb1964-95eb-4d1a-9961-8cf3fe9961e4",
165
- "metadata": {},
166
- "outputs": [],
167
- "source": [
168
- "import pandas as pd\n",
169
- "import matplotlib.pyplot as plt\n",
170
- "from sklearn.preprocessing import StandardScaler\n",
171
- "from sklearn.model_selection import train_test_split\n",
172
- "from sklearn.discriminant_analysis import LinearDiscriminantAnalysis as LDA\n",
173
- "from sklearn.metrics import accuracy_score\n",
174
- "\n",
175
- "df = pd.read_csv(r\"data\\doctor-visits.csv\")\n",
176
- "print(\"Shape:\", df.shape)\n",
177
- "display(df.head())\n",
178
- "\n",
179
- "X = df.drop(columns=['Number of Doctors Visited'])\n",
180
- "y = df['Number of Doctors Visited']\n",
181
- "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.4, \n",
182
- " random_state=4)\n",
183
- "\n",
184
- "lda = LDA(n_components=2)\n",
185
- "X_train = lda.fit_transform(X_train, y_train)\n",
186
- "X_test = lda.fit_transform(X_test, y_test)\n",
187
- "\n",
188
- "lda.fit(X_train,y_train)\n",
189
- "y_pred = lda.predict(X_test)\n",
190
- "print (\"Accuracy:\", accuracy_score(y_test, y_pred))\n",
191
- "\n",
192
- "plt.scatter(X_train[:, 0], X_train[:, 1], c=y_train, cmap='viridis')\n",
193
- "plt.title('LDA Dimensionality Reduction')\n",
194
- "plt.xlabel('LDA Component 1')\n",
195
- "plt.ylabel('LDA Component 2')\n",
196
- "plt.show()"
197
- ]
198
- },
199
- {
200
- "cell_type": "markdown",
201
- "id": "468327ab-beff-425b-ba7a-3e2b373ba9ae",
202
- "metadata": {},
203
- "source": [
204
- "### __***Linear Regression***__"
205
- ]
206
- },
207
- {
208
- "cell_type": "code",
209
- "execution_count": null,
210
- "id": "2511f5c4-b720-4889-b1f3-ca92266bf09b",
211
- "metadata": {},
212
- "outputs": [],
213
- "source": [
214
- "import pandas as pd\n",
215
- "import matplotlib.pyplot as plt\n",
216
- "from sklearn.model_selection import train_test_split\n",
217
- "from sklearn.linear_model import LinearRegression\n",
218
- "from sklearn.metrics import r2_score, mean_squared_error\n",
219
- "\n",
220
- "df = pd.read_excel(\"data/real-estate.xlsx\")\n",
221
- "print(\"Shape:\", df.shape)\n",
222
- "display(df.head())\n",
223
- "\n",
224
- "X = df[['X5 latitude']].values\n",
225
- "y = df['Y house price of unit area'].values\n",
226
- "\n",
227
- "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)\n",
228
- "\n",
229
- "model = LinearRegression().fit(X_train, y_train)\n",
230
- "y_train_pred = model.predict(X_train)\n",
231
- "y_test_pred = model.predict(X_test)\n",
232
- "\n",
233
- "sns.heatmap(df.corr(), annot=True)\n",
234
- "plt.show()\n",
235
- "\n",
236
- "plt.figure(figsize=(10,5))\n",
237
- "plt.scatter(X_train, y_train, label='Train Data Points', edgecolor='k')\n",
238
- "plt.scatter(X_test, y_test, label='Test Data Points', edgecolor='k')\n",
239
- "plt.plot(X_train, model.predict(X_train), color='red', \n",
240
- " label='Linear Regression Line')\n",
241
- "plt.legend()\n",
242
- "plt.grid()\n",
243
- "plt.show()\n",
244
- "\n",
245
- "plt.figure(figsize=(10,5))\n",
246
- "plt.scatter(X_test, y_test, label='Test Data Points', edgecolor='k')\n",
247
- "plt.plot(X_test, model.predict(X_test), color='red', \n",
248
- " label='Linear Regression Line')\n",
249
- "for i in range(len(X_test)):\n",
250
- " plt.plot((X_test[i], X_test[i]), (y_test[i], y_test_pred[i]), \n",
251
- " color='blue', linestyle='--')\n",
252
- "plt.legend()\n",
253
- "plt.grid()\n",
254
- "plt.show()\n",
255
- "\n",
256
- "print(\"Train MSE:\", mean_squared_error(y_train, y_train_pred))\n",
257
- "print(\"Train R2 Score:\", r2_score(y_train, y_train_pred))\n",
258
- "print(\"Test MSE:\", mean_squared_error(y_test, y_test_pred))\n",
259
- "print(\"Test R2 Score:\", r2_score(y_test, y_test_pred))"
260
- ]
261
- },
262
- {
263
- "cell_type": "markdown",
264
- "id": "c4a1c9ec-ca2e-483c-961b-2214c6f966d2",
265
- "metadata": {},
266
- "source": [
267
- "### __***Logistic Regression***__"
268
- ]
269
- },
270
- {
271
- "cell_type": "code",
272
- "execution_count": null,
273
- "id": "39fd591b-1bf3-48ad-83df-7a0832fe5742",
274
- "metadata": {},
275
- "outputs": [],
276
- "source": [
277
- "import pandas as pd\n",
278
- "import numpy as np\n",
279
- "import seaborn as sns\n",
280
- "import matplotlib.pyplot as plt\n",
281
- "from sklearn.preprocessing import MinMaxScaler, LabelEncoder\n",
282
- "from sklearn.model_selection import train_test_split\n",
283
- "from sklearn.decomposition import PCA\n",
284
- "from sklearn.linear_model import LogisticRegression\n",
285
- "from sklearn.metrics import accuracy_score, classification_report, auc\n",
286
- "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay, roc_curve\n",
287
- "\n",
288
- "df = pd.read_csv('data/magic04.data', header=None)\n",
289
- "display(df.head())\n",
290
- "\n",
291
- "X = MinMaxScaler().fit_transform(df.drop(columns=[10]))\n",
292
- "X = PCA(n_components=1).fit_transform(X)\n",
293
- "y = LabelEncoder().fit_transform(df[10]) # Convert 'g'/'h' to 0/1\n",
294
- "\n",
295
- "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, \n",
296
- " stratify=y)\n",
297
- "lr = LogisticRegression(max_iter=5000, random_state=0)\n",
298
- "lr.fit(X_train, y_train)\n",
299
- "y_pred = lr.predict(X_test)\n",
300
- "\n",
301
- "print(f\"Accuracy: {accuracy_score(y_test, y_pred)}\")\n",
302
- "print(classification_report(y_test, y_pred))\n",
303
- "cm = confusion_matrix(y_test, y_pred)\n",
304
- "ConfusionMatrixDisplay(cm, display_labels=['g', 'h']).plot()\n",
305
- "plt.show()\n",
306
- "\n",
307
- "y_pred_proba = lr.predict_proba(X_test)[:, 1]\n",
308
- "fpr, tpr, _ = roc_curve(y_test, y_pred_proba, pos_label=1)\n",
309
- "roc_auc = auc(fpr, tpr)\n",
310
- "plt.plot(fpr, tpr, label=f'ROC curve (area = {roc_auc:.2f})')\n",
311
- "plt.plot([0, 1], [0, 1], 'k--', label='No Skill')\n",
312
- "plt.xlabel('False Positive Rate')\n",
313
- "plt.ylabel('True Positive Rate')\n",
314
- "plt.title('ROC Curve')\n",
315
- "plt.legend()\n",
316
- "plt.show()\n",
317
- "\n",
318
- "# may not proper s curve because of low model accuracy\n",
319
- "# ====================================================\n",
320
- "plt.figure(figsize=(10, 6))\n",
321
- "plt.scatter(X, y, color='red', label='Data points (g = 1, h = 0)')\n",
322
- "x_values = np.linspace(X.min(), X.max(), 100).reshape(-1, 1)\n",
323
- "y_values = lr.predict_proba(x_values)[:, 1] \n",
324
- "plt.plot(x_values, y_values, color='blue', label='Logistic Regression S-Curve')\n",
325
- "plt.axhline(0.5, color='green', linestyle='--', label='Threshold (0.5)')\n",
326
- "plt.title('Logistic Regression: X(PCA Transformed) vs Probability of g')\n",
327
- "plt.legend()\n",
328
- "plt.grid()\n",
329
- "plt.show()"
330
- ]
331
- },
332
- {
333
- "cell_type": "markdown",
334
- "id": "9a251e04-2e7b-4a94-921e-9fd3f18984e7",
335
- "metadata": {},
336
- "source": [
337
- "### __***Naive Bayes***__"
338
- ]
339
- },
340
- {
341
- "cell_type": "code",
342
- "execution_count": null,
343
- "id": "985bcacf-bcee-49ef-9c59-ede381fef350",
344
- "metadata": {},
345
- "outputs": [],
346
- "source": [
347
- "import pandas as pd\n",
348
- "import matplotlib.pyplot as plt\n",
349
- "from sklearn.preprocessing import LabelEncoder\n",
350
- "from sklearn.model_selection import train_test_split\n",
351
- "from sklearn.naive_bayes import GaussianNB\n",
352
- "from sklearn.metrics import accuracy_score, classification_report, roc_curve\n",
353
- "from sklearn.metrics import ConfusionMatrixDisplay, auc, confusion_matrix\n",
354
- "\n",
355
- "df = pd.read_csv(\"data/agaricus-lepiota.data\", header=None)\n",
356
- "display(df.head())\n",
357
- "\n",
358
- "df = df.apply(LabelEncoder().fit_transform)\n",
359
- "X = df.drop(columns=[0])\n",
360
- "y = df[0]\n",
361
- "\n",
362
- "X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=42, \n",
363
- " stratify=y)\n",
364
- "\n",
365
- "nb = GaussianNB()\n",
366
- "nb.fit(X_train, y_train)\n",
367
- "y_pred = nb.predict(X_test)\n",
368
- "\n",
369
- "print(f\"Accuracy: {accuracy_score(y_test, y_pred)}\")\n",
370
- "print(classification_report(y_test, y_pred))\n",
371
- "ConfusionMatrixDisplay(confusion_matrix(y_test, y_pred)).plot()\n",
372
- "plt.show()\n",
373
- "\n",
374
- "y_pred_proba = nb.predict_proba(X_test)[:, 1]\n",
375
- "fpr, tpr, _ = roc_curve(y_test, y_pred_proba)\n",
376
- "roc_auc = auc(fpr, tpr)\n",
377
- "plt.plot(fpr, tpr, label=f'ROC curve (area = {roc_auc:.2f})')\n",
378
- "plt.plot([0, 1], [0, 1], 'k--', label='No Skill')\n",
379
- "plt.xlabel('False Positive Rate')\n",
380
- "plt.ylabel('True Positive Rate')\n",
381
- "plt.title('ROC Curve for Agaricus-Lepiota Classification')\n",
382
- "plt.legend()\n",
383
- "plt.show()"
384
- ]
385
- },
386
- {
387
- "cell_type": "markdown",
388
- "id": "77d7fd14-037a-4e11-b248-c3a7f136a9fc",
389
- "metadata": {},
390
- "source": [
391
- "### __***SVM (Linear & Non-Linear)***__"
392
- ]
393
- },
394
- {
395
- "cell_type": "code",
396
- "execution_count": null,
397
- "id": "17f1e639-d8ae-41f9-9ba7-1c489e122ff0",
398
- "metadata": {},
399
- "outputs": [],
400
- "source": [
401
- "import pandas as pd\n",
402
- "import matplotlib.pyplot as plt\n",
403
- "from sklearn.model_selection import train_test_split\n",
404
- "from sklearn.metrics import accuracy_score, classification_report\n",
405
- "from sklearn.metrics import ConfusionMatrixDisplay, roc_curve, auc\n",
406
- "from sklearn.preprocessing import StandardScaler\n",
407
- "from sklearn.decomposition import PCA\n",
408
- "from sklearn.svm import SVC\n",
409
- "from mlxtend.plotting import plot_decision_regions\n",
410
- "from scipy.io import arff\n",
411
- "\n",
412
- "data = pd.DataFrame(arff.loadarff(\"data/rice.arff\")[0])\n",
413
- "data['Class'] = data['Class'].map({b'Cammeo': 0, b'Osmancik': 1})\n",
414
- "\n",
415
- "X = PCA(n_components=2).fit_transform(data.drop('Class', axis=1))\n",
416
- "y = data['Class']\n",
417
- "\n",
418
- "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, \n",
419
- " random_state=42)\n",
420
- "\n",
421
- "scaler = StandardScaler()\n",
422
- "X_train, X_test = scaler.fit_transform(X_train), scaler.transform(X_test)\n",
423
- "\n",
424
- "# kernel options: 'linear', 'rbf', 'poly', 'sigmoid'\n",
425
- "kernel = 'linear'\n",
426
- "\n",
427
- "# gamma should be:\n",
428
- "# - 'auto' for linear kernel\n",
429
- "# - 'scale' or 'auto' for rbf, poly, sigmoid kernels\n",
430
- "model = SVC(kernel=kernel, C=1, degree=5, \n",
431
- " gamma='auto' if kernel == 'linear' else 'scale', probability=True)\n",
432
- "model.fit(X_train, y_train)\n",
433
- "y_pred = model.predict(X_test)\n",
434
- "\n",
435
- "print(f\"\\nSVM ({kernel}) Accuracy: {accuracy_score(y_test, y_pred):.2f}\")\n",
436
- "print(classification_report(y_test, y_pred))\n",
437
- "\n",
438
- "plot_decision_regions(X_train, y_train.values, clf=model, legend=2)\n",
439
- "plt.title(f'Decision Boundary ({kernel})')\n",
440
- "plt.show()\n",
441
- "\n",
442
- "ConfusionMatrixDisplay.from_predictions(y_test, y_pred, display_labels=[0, 1])\n",
443
- "plt.title(f'Confusion Matrix ({kernel})')\n",
444
- "plt.show()\n",
445
- "\n",
446
- "proba = model.predict_proba(X_test)[:, 1]\n",
447
- "fpr, tpr, _ = roc_curve(y_test, proba)\n",
448
- "plt.plot(fpr, tpr, label=f'{kernel} (AUC = {auc(fpr, tpr):.2f})')\n",
449
- "plt.plot([0, 1], [0, 1], 'k--')\n",
450
- "plt.xlabel('False Positive Rate')\n",
451
- "plt.ylabel('True Positive Rate')\n",
452
- "plt.title(f'ROC Curve ({kernel})')\n",
453
- "plt.legend()\n",
454
- "plt.show()"
455
- ]
456
- },
457
- {
458
- "cell_type": "markdown",
459
- "id": "3047d014-1588-4ba3-b122-8a1e009db945",
460
- "metadata": {},
461
- "source": [
462
- "### __***Feed Forward - Classification (Output Categorical)***__"
463
- ]
464
- },
465
- {
466
- "cell_type": "code",
467
- "execution_count": null,
468
- "id": "02cd39c1-edaa-4f18-9298-68e111cf45ce",
469
- "metadata": {},
470
- "outputs": [],
471
- "source": [
472
- "import warnings\n",
473
- "warnings.filterwarnings('ignore')"
474
- ]
475
- },
476
- {
477
- "cell_type": "code",
478
- "execution_count": null,
479
- "id": "8d6b8e4d-e851-4874-af1f-2da2346a9aa3",
480
- "metadata": {},
481
- "outputs": [],
482
- "source": [
483
- "import numpy as np\n",
484
- "import pandas as pd\n",
485
- "import matplotlib.pyplot as plt\n",
486
- "import tensorflow as tf\n",
487
- "from sklearn.model_selection import train_test_split\n",
488
- "from sklearn.metrics import classification_report, accuracy_score \n",
489
- "from sklearn.metrics import ConfusionMatrixDisplay, roc_curve, auc\n",
490
- "from tensorflow.keras.models import Sequential\n",
491
- "from tensorflow.keras.layers import Dense, Dropout\n",
492
- "from tensorflow.keras.optimizers import Adam\n",
493
- "from tensorflow.keras import regularizers\n",
494
- "\n",
495
- "df = pd.read_csv(\"data/sobar-72.csv\")\n",
496
- "print(\"Shape:\",df.shape)\n",
497
- "display(df.head())\n",
498
- "\n",
499
- "X = df.drop(columns=['ca_cervix'])\n",
500
- "y = tf.keras.utils.to_categorical(df['ca_cervix'], num_classes=2)\n",
501
- "\n",
502
- "x_train, x_test, y_train, y_test = train_test_split(X, y, test_size=0.2, \n",
503
- " random_state=42)\n",
504
- "\n",
505
- "# change l2 (Ridge) to l1 for Lasso regularization\n",
506
- "model = Sequential([\n",
507
- " Dense(128, activation='relu', input_shape=(19,), \n",
508
- " kernel_regularizer=regularizers.l2(0.001)), # if no regularizer remove this\n",
509
- " Dropout(0.3),\n",
510
- " Dense(64, activation='relu', \n",
511
- " kernel_regularizer=regularizers.l2(0.001)), # if no regularizer remove this\n",
512
- " Dropout(0.3),\n",
513
- " Dense(2, activation='softmax') # makes it classification\n",
514
- "])\n",
515
- "\n",
516
- "model.compile(optimizer=Adam(), loss='categorical_crossentropy', \n",
517
- " metrics=['accuracy'])\n",
518
- "\n",
519
- "history = model.fit(x_train, y_train, epochs=50, batch_size=32, \n",
520
- " validation_split=0.2, verbose=1)\n",
521
- "\n",
522
- "def plot_history(history):\n",
523
- " plt.figure(figsize=(12, 5))\n",
524
- " plt.subplot(1, 2, 1)\n",
525
- " plt.plot(history.history['loss'], label='Training Loss')\n",
526
- " plt.plot(history.history['val_loss'], label='Validation Loss')\n",
527
- " plt.title('Loss')\n",
528
- " plt.subplot(1, 2, 2)\n",
529
- " plt.plot(history.history['accuracy'], label='Training Accuracy')\n",
530
- " plt.plot(history.history['val_accuracy'], label='Validation Accuracy')\n",
531
- " plt.title('Accuracy')\n",
532
- " plt.tight_layout()\n",
533
- " plt.show()\n",
534
- "\n",
535
- "plot_history(history)\n",
536
- "\n",
537
- "loss, accuracy = model.evaluate(x_test, y_test)\n",
538
- "print(f'Test accuracy: {accuracy:.4f}, Test loss: {loss:.4f}')\n",
539
- "\n",
540
- "y_pred = np.argmax(model.predict(x_test), axis=1)\n",
541
- "y_test = np.argmax(y_test, axis=1)\n",
542
- "\n",
543
- "print(\"Classification Report:\")\n",
544
- "print(classification_report(y_test, y_pred))\n",
545
- "print(f\"Accuracy: {accuracy_score(y_test, y_pred):.2f}\")\n",
546
- "ConfusionMatrixDisplay.from_predictions(y_test, y_pred, cmap='Blues')\n",
547
- "plt.show()\n",
548
- "\n",
549
- "y_pred_proba = model.predict(x_test)[:, 1]\n",
550
- "fpr, tpr, _ = roc_curve(y_test, y_pred_proba)\n",
551
- "plt.plot(fpr, tpr, label=f'ROC curve (area = {auc(fpr, tpr):.2f})')\n",
552
- "plt.plot([0, 1], [0, 1], color='navy', linestyle='--')\n",
553
- "plt.xlabel('False Positive Rate')\n",
554
- "plt.ylabel('True Positive Rate')\n",
555
- "plt.title('Receiver Operating Characteristic')\n",
556
- "plt.legend(loc=\"lower right\")\n",
557
- "plt.show()"
558
- ]
559
- },
560
- {
561
- "cell_type": "markdown",
562
- "id": "86c71922-b7d4-4ae8-ae4e-0f19b4b74af7",
563
- "metadata": {},
564
- "source": [
565
- "### __***Feed Forward - Regression (Output Numerical)***__"
566
- ]
567
- },
568
- {
569
- "cell_type": "code",
570
- "execution_count": null,
571
- "id": "ac35d507-57d3-4ea6-a535-3df21c0b4dcd",
572
- "metadata": {},
573
- "outputs": [],
574
- "source": [
575
- "import warnings\n",
576
- "warnings.filterwarnings('ignore')"
577
- ]
578
- },
579
- {
580
- "cell_type": "code",
581
- "execution_count": null,
582
- "id": "02e338ba-2fa5-4187-8337-a4db19af2faf",
583
- "metadata": {},
584
- "outputs": [],
585
- "source": [
586
- "import numpy as np\n",
587
- "import matplotlib.pyplot as plt\n",
588
- "from sklearn.model_selection import train_test_split\n",
589
- "from sklearn.preprocessing import StandardScaler\n",
590
- "from sklearn.metrics import accuracy_score, mean_squared_error\n",
591
- "import tensorflow as tf\n",
592
- "from tensorflow.keras.models import Sequential\n",
593
- "from tensorflow.keras.layers import Dense\n",
594
- "\n",
595
- "df = pd.read_csv('data/california.csv')\n",
596
- "display(df.head())\n",
597
- "\n",
598
- "X = df.drop(columns='target')\n",
599
- "y = df['target']\n",
600
- "\n",
601
- "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, \n",
602
- " random_state=42)\n",
603
- "\n",
604
- "scaler = StandardScaler()\n",
605
- "X_train = scaler.fit_transform(X_train)\n",
606
- "X_test = scaler.transform(X_test)\n",
607
- "\n",
608
- "# Simple FNN/MLP model\n",
609
- "model = Sequential([\n",
610
- " Dense(32, activation='relu', input_shape=(X_train.shape[1],)),\n",
611
- " Dense(16, activation='relu'),\n",
612
- " Dense(1) # makes its regression\n",
613
- "])\n",
614
- "\n",
615
- "model.compile(optimizer='adam', loss='mse')\n",
616
- "\n",
617
- "history = model.fit(X_train, y_train, epochs=25, validation_split=0.2, verbose=1)\n",
618
- "\n",
619
- "y_pred = model.predict(X_test).flatten()\n",
620
- "mse = mean_squared_error(y_test, y_pred)\n",
621
- "print(f\"Regression MSE on California Housing dataset: {mse:.4f}\")\n",
622
- "\n",
623
- "plt.plot(history.history['loss'], label='Train Loss')\n",
624
- "plt.plot(history.history['val_loss'], label='Val Loss')\n",
625
- "plt.title('Regression Loss')\n",
626
- "plt.xlabel('Epochs')\n",
627
- "plt.ylabel('MSE Loss')\n",
628
- "plt.legend()\n",
629
- "\n",
630
- "plt.tight_layout()\n",
631
- "plt.show()"
632
- ]
633
- },
634
- {
635
- "cell_type": "markdown",
636
- "id": "5253df17-9602-4e1e-adec-df37b95a87f0",
637
- "metadata": {},
638
- "source": [
639
- "### __***MLP (FNN from scratch above code or learn this)***__"
640
- ]
641
- },
642
- {
643
- "cell_type": "code",
644
- "execution_count": null,
645
- "id": "886ac28f-4952-4a26-bd2f-912cac7e8107",
646
- "metadata": {},
647
- "outputs": [],
648
- "source": [
649
- "import pandas as pd\n",
650
- "from sklearn.preprocessing import LabelEncoder\n",
651
- "from sklearn.model_selection import train_test_split\n",
652
- "from sklearn.metrics import classification_report, confusion_matrix\n",
653
- "from sklearn.metrics import ConfusionMatrixDisplay\n",
654
- "from sklearn.neural_network import MLPClassifier\n",
655
- "\n",
656
- "df = pd.read_csv('data/HeartDiseaseTrain-Test.csv')\n",
657
- "display(df.head())\n",
658
- "\n",
659
- "X = df.drop('target', axis=1)\n",
660
- "X = X.apply(LabelEncoder().fit_transform)\n",
661
- "\n",
662
- "X = X / X.max() #normalize features\n",
663
- "y = df['target']\n",
664
- "\n",
665
- "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, \n",
666
- " random_state=1)\n",
667
- "\n",
668
- "# Using 'relu' activation; others to try: 'identity', 'tanh', 'logistic'\n",
669
- "activation = 'relu'\n",
670
- "hidden_layers = (8, 8, 8)\n",
671
- "\n",
672
- "print(f\"\\nActivation: {activation}, Hidden Layers: {hidden_layers}\")\n",
673
- "model = MLPClassifier(hidden_layer_sizes=hidden_layers, activation=activation,\n",
674
- " solver='adam', max_iter=500, random_state=42)\n",
675
- "model.fit(X_train, y_train)\n",
676
- "\n",
677
- "preds_train = model.predict(X_train)\n",
678
- "print(\"\\nTrain Results:\")\n",
679
- "print(classification_report(y_train, preds_train, zero_division=0))\n",
680
- "ConfusionMatrixDisplay.from_predictions(y_train, preds_train)\n",
681
- "plt.show()\n",
682
- "\n",
683
- "preds_test = model.predict(X_test)\n",
684
- "print(\"\\nTest Results:\")\n",
685
- "print(classification_report(y_test, preds_test, zero_division=0))\n",
686
- "ConfusionMatrixDisplay.from_predictions(y_test, preds_test)\n",
687
- "plt.show()"
688
- ]
689
- },
690
- {
691
- "cell_type": "markdown",
692
- "id": "625ac47b-8d21-4e64-9729-791067f4edca",
693
- "metadata": {},
694
- "source": [
695
- "### __***CNN***__"
696
- ]
697
- },
698
- {
699
- "cell_type": "code",
700
- "execution_count": null,
701
- "id": "6e34cf42-1b2b-49a9-81c0-c66e380122b0",
702
- "metadata": {},
703
- "outputs": [],
704
- "source": [
705
- "import warnings\n",
706
- "warnings.filterwarnings('ignore')"
707
- ]
708
- },
709
- {
710
- "cell_type": "code",
711
- "execution_count": null,
712
- "id": "9d1d0dc8-af7c-4052-8834-936143fb490c",
713
- "metadata": {},
714
- "outputs": [],
715
- "source": [
716
- "import tensorflow as tf\n",
717
- "from tensorflow.keras import layers, models\n",
718
- "from keras.datasets import cifar10\n",
719
- "from keras.utils import to_categorical\n",
720
- "import matplotlib.pyplot as plt\n",
721
- "from sklearn.metrics import roc_curve, auc, confusion_matrix\n",
722
- "from sklearn.metrics import ConfusionMatrixDisplay\n",
723
- "from sklearn.preprocessing import label_binarize\n",
724
- "\n",
725
- "(X_train, y_train), (X_test, y_test) = cifar10.load_data()\n",
726
- "\n",
727
- "X_train = X_train.astype('float32') / 255.0\n",
728
- "X_test = X_test.astype('float32') / 255.0\n",
729
- "\n",
730
- "y_train = to_categorical(y_train, 10)\n",
731
- "y_test = to_categorical(y_test, 10)\n",
732
- "\n",
733
- "model = models.Sequential([\n",
734
- " layers.Conv2D(32, kernel_size=(3, 3), activation=\"relu\", input_shape=(32, 32, 3)),\n",
735
- " layers.MaxPooling2D(),\n",
736
- " layers.Conv2D(64, kernel_size=(3, 3), activation=\"relu\"),\n",
737
- " layers.MaxPooling2D(),\n",
738
- " layers.Flatten(),\n",
739
- " layers.Dense(10, activation=\"softmax\")\n",
740
- "])\n",
741
- "\n",
742
- "model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])\n",
743
- "\n",
744
- "history = model.fit(X_train, y_train, epochs=5, batch_size=32, validation_split=0.2)\n",
745
- "\n",
746
- "y_pred = model.predict(X_test)\n",
747
- "y_pred_classes = tf.argmax(y_pred, axis=1)\n",
748
- "y_test_classes = tf.argmax(y_test, axis=1)\n",
749
- "\n",
750
- "conf_matrix = confusion_matrix(y_test_classes, y_pred_classes)\n",
751
- "ConfusionMatrixDisplay(conf_matrix).plot(cmap='Blues')\n",
752
- "plt.title('Confusion Matrix')\n",
753
- "plt.show()\n",
754
- "\n",
755
- "plt.figure(figsize=(12, 5))\n",
756
- "\n",
757
- "plt.subplot(1, 2, 1)\n",
758
- "plt.plot(history.history['accuracy'], label='Training Accuracy')\n",
759
- "plt.plot(history.history['val_accuracy'], label='Validation Accuracy')\n",
760
- "plt.title('Accuracy Curve')\n",
761
- "plt.xlabel('Epochs')\n",
762
- "plt.ylabel('Accuracy')\n",
763
- "plt.legend()\n",
764
- "\n",
765
- "plt.subplot(1, 2, 2)\n",
766
- "plt.plot(history.history['loss'], label='Training Loss')\n",
767
- "plt.plot(history.history['val_loss'], label='Validation Loss')\n",
768
- "plt.title('Loss Curve')\n",
769
- "plt.xlabel('Epochs')\n",
770
- "plt.ylabel('Loss')\n",
771
- "plt.legend()\n",
772
- "\n",
773
- "plt.tight_layout()\n",
774
- "plt.show()"
775
- ]
776
- },
777
- {
778
- "cell_type": "markdown",
779
- "id": "af7093fd-ea5c-4058-b4a3-3692a5e61eff",
780
- "metadata": {},
781
- "source": [
782
- "### __***CNN Another Example***__"
783
- ]
784
- },
785
- {
786
- "cell_type": "code",
787
- "execution_count": null,
788
- "id": "381a98d6-7828-43c8-b0a4-d1c24b229e66",
789
- "metadata": {},
790
- "outputs": [],
791
- "source": [
792
- "import numpy as np\n",
793
- "import tensorflow as tf\n",
794
- "from tensorflow.keras import layers, models\n",
795
- "from tensorflow.keras.preprocessing import image\n",
796
- "import os\n",
797
- "\n",
798
- "dataset_path = 'Pistachio_Image_Dataset'\n",
799
- "\n",
800
- "img_height, img_width = 180, 180\n",
801
- "batch_size = 32\n",
802
- "\n",
803
- "train_ds = tf.keras.utils.image_dataset_from_directory(\n",
804
- " dataset_path,\n",
805
- " validation_split=0.2,\n",
806
- " subset=\"training\",\n",
807
- " seed=123,\n",
808
- " image_size=(img_height, img_width),\n",
809
- " batch_size=batch_size)\n",
810
- "\n",
811
- "val_ds = tf.keras.utils.image_dataset_from_directory(\n",
812
- " dataset_path,\n",
813
- " validation_split=0.2,\n",
814
- " subset=\"validation\",\n",
815
- " seed=123,\n",
816
- " image_size=(img_height, img_width),\n",
817
- " batch_size=batch_size)\n",
818
- "\n",
819
- "model = models.Sequential([\n",
820
- " layers.Rescaling(1./255, input_shape=(img_height, img_width, 3)),\n",
821
- " layers.Conv2D(32, 3, activation='relu'),\n",
822
- " layers.MaxPooling2D(),\n",
823
- " layers.Conv2D(64, 3, activation='relu'),\n",
824
- " layers.MaxPooling2D(),\n",
825
- " layers.Conv2D(128, 3, activation='relu'),\n",
826
- " layers.MaxPooling2D(),\n",
827
- " layers.Flatten(),\n",
828
- " layers.Dense(128, activation='relu'),\n",
829
- " layers.Dense(2)\n",
830
- "])\n",
831
- "\n",
832
- "model.compile(optimizer='adam',\n",
833
- " loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
834
- " metrics=['accuracy'])\n",
835
- "\n",
836
- "epochs = 5\n",
837
- "history = model.fit(\n",
838
- " train_ds,\n",
839
- " validation_data=val_ds,\n",
840
- " epochs=epochs\n",
841
- ")\n",
842
- "\n",
843
- "class_names = ['Kirmizi_Pistachio', 'Siirt_Pistachio']\n",
844
- "img_path = 'Pistachio_Image_Dataset/Siirt_Pistachio/siirt (11).jpg'\n",
845
- "img = image.load_img(img_path, target_size=(180, 180))\n",
846
- "\n",
847
- "plt.imshow(img)\n",
848
- "plt.title(\"Input Image\")\n",
849
- "plt.axis(\"off\")\n",
850
- "plt.show()\n",
851
- "\n",
852
- "img_array = image.img_to_array(img)\n",
853
- "img_array = tf.expand_dims(img_array, 0)\n",
854
- "\n",
855
- "predictions = model.predict(img_array)\n",
856
- "score = tf.nn.softmax(predictions[0])\n",
857
- "\n",
858
- "predicted_class = class_names[np.argmax(score)]\n",
859
- "confidence = 100 * np.max(score)\n",
860
- "\n",
861
- "print(f\"Image most likely belongs to '{predicted_class}' with {confidence:.2f}% confidence.\")"
862
- ]
863
- },
864
- {
865
- "cell_type": "markdown",
866
- "id": "770b6523-d178-45f6-b6a7-a04d780f5a9a",
867
- "metadata": {},
868
- "source": [
869
- "### __***HMM - Viterbi***__"
870
- ]
871
- },
872
- {
873
- "cell_type": "code",
874
- "execution_count": null,
875
- "id": "eebdc5a0-8aef-41ff-ad89-d9424aa8a43f",
876
- "metadata": {},
877
- "outputs": [],
878
- "source": [
879
- "import numpy as np\n",
880
- "import pandas as pd\n",
881
- "import networkx as nx\n",
882
- "import matplotlib.pyplot as plt\n",
883
- "from hmmlearn.hmm import CategoricalHMM\n",
884
- "\n",
885
- "visibleStates = ['early', 'mid', 'late']\n",
886
- "hiddenStates = ['Genuine User', 'Intruder']\n",
887
- "hiddenInitial = [0.9, 0.1]\n",
888
- "\n",
889
- "hiddenTransition = np.array([\n",
890
- " [0.7, 0.3],\n",
891
- " [0.4, 0.6]\n",
892
- "])\n",
893
- "\n",
894
- "emissionMatrix = np.array([\n",
895
- " [0.8, 0.1, 0.1],\n",
896
- " [0.1, 0.3, 0.6]\n",
897
- "])\n",
898
- "\n",
899
- "graph = nx.DiGraph()\n",
900
- "graph.add_nodes_from(visibleStates + hiddenStates)\n",
901
- "\n",
902
- "for i, x in enumerate(hiddenStates):\n",
903
- " for j, y in enumerate(hiddenStates):\n",
904
- " graph.add_edge(x, y, weight=hiddenTransition[i, j])\n",
905
- "\n",
906
- "for i, x in enumerate(hiddenStates):\n",
907
- " for j, y in enumerate(visibleStates):\n",
908
- " graph.add_edge(x, y, weight=emissionMatrix[i, j])\n",
909
- "\n",
910
- "pos = nx.circular_layout(graph)\n",
911
- "nx.draw(graph, pos, with_labels=True, node_size=1500)\n",
912
- "nx.draw_networkx_edge_labels(graph, pos, edge_labels=nx.get_edge_attributes(graph, 'weight'))\n",
913
- "plt.show()\n",
914
- "\n",
915
- "observations = ['early', 'early', 'late', 'mid', 'early', 'late']\n",
916
- "observationMap = {'early': 0, 'mid': 1, 'late': 2}\n",
917
- "mappedSequence = np.array([observationMap[o] for o in observations]).reshape(-1, 1)\n",
918
- "\n",
919
- "model = CategoricalHMM(n_components=2)\n",
920
- "model.startprob_ = hiddenInitial\n",
921
- "model.transmat_ = hiddenTransition\n",
922
- "model.emissionprob_ = emissionMatrix\n",
923
- "\n",
924
- "logValue, bestPath = model.decode(mappedSequence, algorithm=\"viterbi\")\n",
925
- "\n",
926
- "decodedPath = [hiddenStates[state] for state in bestPath]\n",
927
- "\n",
928
- "result = pd.DataFrame({\n",
929
- " 'Observation': observations,\n",
930
- " 'Predicted State': decodedPath\n",
931
- "})\n",
932
- "\n",
933
- "print(\"\\nDecoded Path with Observations:\")\n",
934
- "display(result)\n",
935
- "\n",
936
- "print(\"\\nLog Probability of Best Path:\", logValue)"
937
- ]
938
- }
939
- ],
940
- "metadata": {
941
- "kernelspec": {
942
- "display_name": "NEW-VENV-1",
943
- "language": "python",
944
- "name": "new-venv-1"
945
- },
946
- "language_info": {
947
- "codemirror_mode": {
948
- "name": "ipython",
949
- "version": 3
950
- },
951
- "file_extension": ".py",
952
- "mimetype": "text/x-python",
953
- "name": "python",
954
- "nbconvert_exporter": "python",
955
- "pygments_lexer": "ipython3",
956
- "version": "3.11.5"
957
- }
958
- },
959
- "nbformat": 4,
960
- "nbformat_minor": 5
961
- }