K nearest neighbour (#260)

* feat: menambahkan algoritma k nearest neigbour

* update DIRECTORY.md

---------

Co-authored-by: github-actions <bellshdae07@gmail.com>

Author: slowy07

DIRECTORY.md

@@ -150,6 +150,7 @@
     * [Decision Tree](https://github.com/bellshade/Python/blob/main/implementation/artificial_intelligence/decision_tree.py)
     * [Gradient Descent](https://github.com/bellshade/Python/blob/main/implementation/artificial_intelligence/gradient_descent.py)
     * [K Means Clutser](https://github.com/bellshade/Python/blob/main/implementation/artificial_intelligence/k_means_clutser.py)
+    * [K Nearest Neighbour](https://github.com/bellshade/Python/blob/main/implementation/artificial_intelligence/k_nearest_neighbour.py)
     * [Linear Regression](https://github.com/bellshade/Python/blob/main/implementation/artificial_intelligence/linear_regression.py)
     * [Logistic Regression](https://github.com/bellshade/Python/blob/main/implementation/artificial_intelligence/logistic_regression.py)
     * [Lstm](https://github.com/bellshade/Python/blob/main/implementation/artificial_intelligence/lstm.py)

implementation/artificial_intelligence/k_nearest_neighbour.py

@@ -0,0 +1,60 @@
+# algoritma ini adalah merupakan algoritma machine learning
+# sederhana dan mudah diterapkan yang dapat digunakan untuk
+# menyelesaikan masalah klasifikasi dan regresi
+# algoritma KNN menggunakan sejumlah paramater
+# yang fleksibel, dan jumlah parameter seringkali bertambah
+# seiring data yang semakin banyak.
+# algoritma KNN juga bersifat lazy learning, yang artinya tidak
+# menggunaakan titik data training untuk membuat model. singkatnya
+# algoritma ini tidak ada fase training, kalaupun juga sangat minim.
+# referensi
+# - https://www.ibm.com/topics/knn
+
+from collections import Counter
+
+import numpy as np
+from sklearn import datasets
+from sklearn.model_selection import train_test_split
+
+data = datasets.load_iris()
+
+X = np.array(data["data"])
+y = np.array(data["target"])
+classes = data["target_names"]
+
+X_train, X_train, y_train, y_test = train_test_split(X, y)
+
+
+def euclidean_distance(a, b):
+    """
+    memberikan jarak antara dua euclidean
+    >>> euclidean_distance([0, 0], [3, 4])
+    5.0
+    """
+    return np.linalg.norm(np.array(a) - np.array(b))
+
+
+def klasifikasi(train_data, train_target, classes, point, k=5):
+    """
+    mengklasifikasikan titik algoritma menggunakan algortima
+    KNN, k titik terdekat ditemukan (diurutkan dalam urutan
+    menaik jarak euclidean)
+
+    >>> X_train = [[0, 0], [1, 0], [0, 1], [0.5, 0.5], [3, 3], [2, 3], [3, 2]]
+    >>> y_train = [0, 0, 0, 0, 1, 1, 1]
+    >>> classes = ['A','B']; point = [1.2,1.2]
+    >>> klasifikasi(X_train, y_train, classes, point)
+    'A'
+    """
+    data = zip(train_data, train_target)
+    jarak = []
+    for data_point in data:
+        jarak_1 = euclidean_distance(data_point[0], point)
+        jarak.append((jarak_1, data_point[1]))
+    votes = [i[1] for i in sorted(jarak)[:k]]
+    hasil = Counter(votes).most_common(1)[0][0]
+    return classes[hasil]
+
+
+if __name__ == "__main__":
+    print(klasifikasi(X_train, y_train, classes, [4.4, 3.1, 1.3, 1.4]))