机器学习学习中-->>>knn手动实现
knn学习中，手动实现测试版 knn思想
1 # 假设test_data1 为类型1 test_data2 为类型2 test_data3 为类型3 2 test_data1 = [(1, 2), (2, 5), (3, 3), (5, 9), (6, 8), (8, 3), (4, 3)] 3 test_data2 = [(15, 6), (2, 6), (8, 6), (3, 1), (4, 5), (2, 1), (3, 6)] 4 test_data3 = [(25, 4), (3, 2), (8, 4), (2, 53), (6, 18), (13, 3), (25, 8)] 5 6 7 def knn(k:int, new_data: tuple, *test_data:list): 8 """ 9 实现knn分类算法 10 :param k: knn中 k的值 11 :param new_data: 新纪录 12 :return: 返回所属类型 13 """ 14 # 用于生成字典的 key, 其中 15 # i = 1 16 j = 1 17 # 用于存储所有的字典 18 all_distance_dict = {} 19 for data in test_data: 20 distance = computed_range(new_data = new_data, test_data = data) 21 if j == 1 : 22 temp = [("j" + str(a)) for a in range(len(distance))] 23 elif j == 2: 24 temp = [("k" + str(a)) for a in range(len(distance))] 25 elif j == 3: 26 temp = [("l" + str(a)) for a in range(len(distance))] 27 28 distance_dict = dict(zip(temp, distance)) 29 all_distance_dict.update(distance_dict) 30 # i += 1 31 j += 1 32 33 34 # 排序后的结果 35 sort_all_distance_list = sorted(all_distance_dict.items(), key=lambda x: x[1], reverse=False) 36 sort_all_distance_dict = dict(sort_all_distance_list) 37 38 print(sort_all_distance_dict) 39 40 # 排序后取前k个 41 end_distance_list = [] 42 # 引入一个计数器 43 i = 0 44 for key in sort_all_distance_dict.keys(): 45 if i != k: 46 end_distance_list.append(key) 47 else : 48 break 49 i += 1 50 51 # 创建计数器 res1 res2 res3 分别表示每个种类的个数 52 res1, res2 ,res3 = 0, 0, 0 53 for key in end_distance_list: 54 if key[0] == "j": 55 res1 += 1 56 elif key[0] =="k" : 57 res2 +=1 58 elif key[0] == "l": 59 res3 += 1 60 61 if res1>res2: 62 if res1>res3: 63 return "类型一" 64 elif res3 > res2 : 65 return "类型三" 66 else : 67 return "类型二" 68 69 70 71 72 # 计算距离 73 def computed_range(new_data: tuple, test_data:list, formula_mode = 1)->list : 74 ''' 75 该函数用于计算欧氏距离 76 :param new_data:新需要计算的数据 77 :param test_data:为样本数据 78 :param formula_mode:用于选择相似度计算方式：其中 1：欧氏距离 2：曼哈顿距离 3：余弦相似度 79 :return:list 代表每个数据与新记录之间的距离 80 ''' 81 result = [] # 定义一个列表：用于存出结果 82 if formula_mode == 1 : 83 formula = Euclidean_distance 84 elif formula_mode == 2 : 85 formula = Manhattan_distance 86 elif formula_mode == 3: 87 formula = cosine_measure 88 else: 89 return # 输入非 1 2 3 则直接返回None 90 91 for data in test_data : 92 # 用于计算距离欧式距离 93 length = formula(data, new_data) 94 result.append(length) 95 96 return result 97 98 # 用于计算欧氏距离 99 def Euclidean_distance(data1,data2) : 100 return (((data1[0] - data2[0]) ** 2) + ((data1[1] - data2[1]) ** 2)) ** (0.5) 101 102 # 用于计算曼哈顿距离 103 def Manhattan_distance() : 104 pass 105 106 # 用于计算余弦相似度 107 def cosine_measure() : 108 pass 109 110 string = knn( 3, (5,55), test_data1, test_data2, test_data3) 111 print(string)
返回结果
{'l3': 3.605551275463989, 'l4': 37.013511046643494, 'j3': 46.0, 'j4': 47.01063709417264, 'k6': 49.040799340956916, 'k1': 49.09175083453431, 'k2': 49.09175083453431, 'k0': 50.00999900019995, 'k4': 50.00999900019995, 'j1': 50.08991914547278, 'l6': 51.07837115648854, 'l2': 51.088159097779204, 'j6': 52.009614495783374, 'j2': 52.03844732503075, 'j5': 52.08646657242167, 'l5': 52.61178575186362, 'l1': 53.03772242470448, 'j0': 53.150729063673246, 'k3': 54.037024344425184, 'k5': 54.08326913195984, 'l0': 54.78138369920935}
类型三
