Spaces:
Sleeping
Sleeping
Shorba
Browse files- KNNMulti.py +30 -0
- __pycache__/KNNMulti.cpython-313.pyc +0 -0
- __pycache__/app.cpython-313.pyc +0 -0
- app.py +4 -5
KNNMulti.py
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import numpy as np
|
| 2 |
+
from collections import Counter
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
class CustomKNN:
|
| 6 |
+
def __init__(self, k=5, weights="distance"):
|
| 7 |
+
self.k = k
|
| 8 |
+
self.weights = weights
|
| 9 |
+
|
| 10 |
+
def fit(self, X, y):
|
| 11 |
+
self.X_train = np.array(X)
|
| 12 |
+
self.y_train = np.array(y)
|
| 13 |
+
|
| 14 |
+
def predict(self, X_test):
|
| 15 |
+
return np.array([self._predict_single(x) for x in np.array(X_test)])
|
| 16 |
+
|
| 17 |
+
def _predict_single(self, x_test):
|
| 18 |
+
distances = np.sqrt(np.sum((self.X_train - x_test) ** 2, axis=1))
|
| 19 |
+
k_indices = np.argsort(distances)[:self.k]
|
| 20 |
+
k_nearest_labels = self.y_train[k_indices]
|
| 21 |
+
|
| 22 |
+
if self.weights == "distance":
|
| 23 |
+
k_distances = distances[k_indices]
|
| 24 |
+
vote_weights = 1.0 / (k_distances + 1e-5)
|
| 25 |
+
class_votes = {}
|
| 26 |
+
for label, weight in zip(k_nearest_labels, vote_weights):
|
| 27 |
+
class_votes[label] = class_votes.get(label, 0) + weight
|
| 28 |
+
return max(class_votes, key=class_votes.get)
|
| 29 |
+
|
| 30 |
+
return Counter(k_nearest_labels).most_common(1)[0][0]
|
__pycache__/KNNMulti.cpython-313.pyc
ADDED
|
Binary file (2.3 kB). View file
|
|
|
__pycache__/app.cpython-313.pyc
ADDED
|
Binary file (10.2 kB). View file
|
|
|
app.py
CHANGED
|
@@ -3,7 +3,7 @@ import numpy as np
|
|
| 3 |
from skimage.feature import hog
|
| 4 |
import cv2
|
| 5 |
import pickle
|
| 6 |
-
from
|
| 7 |
|
| 8 |
|
| 9 |
|
|
@@ -31,10 +31,9 @@ def build_knn_from_pca_package(package):
|
|
| 31 |
print(f"knn_pca.pkl is missing keys: {sorted(missing_keys)}")
|
| 32 |
return None
|
| 33 |
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
)
|
| 38 |
|
| 39 |
gnb_package = load_pickle_file("gnb_digit6_package.pkl")
|
| 40 |
pca_multi = load_pickle_file("pcaMulti.pkl")
|
|
|
|
| 3 |
from skimage.feature import hog
|
| 4 |
import cv2
|
| 5 |
import pickle
|
| 6 |
+
from KNNMulti import CustomKNN as MultimodeKNN
|
| 7 |
|
| 8 |
|
| 9 |
|
|
|
|
| 31 |
print(f"knn_pca.pkl is missing keys: {sorted(missing_keys)}")
|
| 32 |
return None
|
| 33 |
|
| 34 |
+
knn = MultimodeKNN(k=package["best_k"], weights=package["weights"])
|
| 35 |
+
knn.fit(package["X_train"], package["y_train"])
|
| 36 |
+
return knn
|
|
|
|
| 37 |
|
| 38 |
gnb_package = load_pickle_file("gnb_digit6_package.pkl")
|
| 39 |
pca_multi = load_pickle_file("pcaMulti.pkl")
|