【数据挖掘与商务智能决策】第十三章数据降维之PCA 主成分分析

13.1.2 PCA主成分分析代码实现

1.二维空间降维Python代码实现

import numpy as np
X = np.array([[1, 1], [2, 2], [3, 3]])
X

array([[1, 1],[2, 2],[3, 3]])

# 也可以通过pandas库来构造数据，效果一样
import pandas as pd
X = pd.DataFrame([[1, 1], [2, 2], [3, 3]])
X

	0	1
0	1	1
1	2	2
2	3	3

# 数据降维，由二维降至一维
from sklearn.decomposition import PCA
pca = PCA(n_components=1)
pca.fit(X)  # 进行降维模型训练
X_transformed = pca.transform(X)  # 进行数据降维，并赋值给X_transformedX_transformed  # 查看降维后的结果

array([[-1.41421356],[ 0.        ],[ 1.41421356]])

# 查看此时的维度
X_transformed.shape

(3, 1)

# 查看降维的系数
pca.components_

array([[0.70710678, 0.70710678]])

# 查看线性组合表达式
a = pca.components_[0][0] 
b = pca.components_[0][1]
print(str(a) + ' * X + ' +  str(b) + ' * Y')

0.7071067811865476 * X + 0.7071067811865475 * Y

这个的确也和12.1.1节我们获得的线性组合是一样的。

2.三维空间降维Python代码实现

import pandas as pd
X = pd.DataFrame([[45, 0.8, 9120], [40, 0.12, 2600], [38, 0.09, 3042], [30, 0.04, 3300], [39, 0.21, 3500]], columns=['年龄(岁)', '负债比率', '月收入(元)'])
X

	年龄(岁)	负债比率	月收入(元)
0	45	0.80	9120
1	40	0.12	2600
2	38	0.09	3042
3	30	0.04	3300
4	39	0.21	3500

# 因为三个指标数据的量级相差较大，所以可以先进行数据归一化处理
from sklearn.preprocessing import StandardScaler
X_new = StandardScaler().fit_transform(X)X_new  # 查看归一化后的数据

array([[ 1.36321743,  1.96044639,  1.98450514],[ 0.33047695, -0.47222431, -0.70685302],[-0.08261924, -0.57954802, -0.52440206],[-1.73500401, -0.75842087, -0.41790353],[ 0.12392886, -0.15025319, -0.33534653]])

from sklearn.decomposition import PCA
pca = PCA(n_components=2)
pca.fit(X_new)  # 进行降维模型训练
X_transformed = pca.transform(X_new)  # 进行数据降维，并赋值给X_transformedX_transformed  # 查看降维后的结果

array([[ 3.08724247,  0.32991205],[-0.52888635, -0.74272137],[-0.70651782, -0.33057258],[-1.62877292,  1.05218639],[-0.22306538, -0.30880449]])

# 查看降维的系数
pca.components_

array([[ 0.52952108,  0.61328179,  0.58608264],[-0.82760701,  0.22182579,  0.51561609]])

dim = ['年龄(岁)', '负债比率', '月收入(元)']
for i in pca.components_:formula = []for j in range(len(i)):formula.append(str(i[j]) + ' * ' + dim[j])print(" + ".join(formula))

0.5295210839165538 * 年龄(岁) + 0.6132817922410683 * 负债比率 + 0.5860826434841948 * 月收入(元)
-0.8276070105929828 * 年龄(岁) + 0.2218257919336094 * 负债比率 + 0.5156160917294705 * 月收入(元)

# 如果不想显示具体的特征名称，可以采用如下的写法
dim = ['X', 'Y', 'Z']
for i in pca.components_:formula = []for j in range(len(i)):formula.append(str(i[j]) + ' * ' + dim[j])print(" + ".join(formula))

0.5295210839165538 * X + 0.6132817922410683 * Y + 0.5860826434841948 * Z
-0.8276070105929828 * X + 0.2218257919336094 * Y + 0.5156160917294705 * Z

13.2 案例实战: 人脸识别模型

13.2.1 案例背景

13.2.2 人脸数据读取、处理与变量提取

1. 读取人脸照片数据

import os
names = os.listdir('olivettifaces')names[0:5]  # 查看前5项读取的文件名

['10_0.jpg', '10_1.jpg', '10_2.jpg', '10_3.jpg', '10_4.jpg']

# 获取到文件名称后，便可以通过如下代码在Python中查看这些图片
from PIL import Image
img0 = Image.open('olivettifaces\\' + names[0])
img0.show()

img0  # 在Jupyter Notebook中可以直接输入变量名查看图像

在这里插入图片描述

2.人脸数据处理: 特征变量提取

# 图像灰度处理及数值化处理
import numpy as np
img0 = img0.convert('L')
img0 = img0.resize((32, 32))
arr = np.array(img0)arr  # 查看数值化后的结果

array([[186,  76,  73, ..., 100, 103, 106],[196,  85,  68, ...,  85, 106, 103],[193,  69,  79, ...,  82,  99, 100],...,[196,  87, 193, ..., 103,  66,  52],[219, 179, 202, ..., 150, 127, 109],[244, 228, 230, ..., 198, 202, 206]], dtype=uint8)

# 如果觉得numpy格式的arr不好观察，则可以通过pandas库将其转为DataFrame格式进行观察
import pandas as pd
pd.DataFrame(arr)

	0	1	2	3	4	5	6	7	8	9	...	22	23	24	25	26	27	28	29	30	31
0	186	76	73	87	89	88	75	81	100	102	...	71	75	75	73	76	85	95	100	103	106
1	196	85	68	78	104	97	100	94	83	87	...	52	59	70	85	62	82	89	85	106	103
2	193	69	79	92	105	102	112	117	106	94	...	41	45	50	76	59	74	83	82	99	100
3	186	67	71	75	85	99	114	115	109	109	...	42	43	40	52	41	61	69	76	76	108
4	179	46	41	50	53	69	80	91	108	104	...	43	37	30	31	35	43	59	61	56	101
5	173	33	43	49	48	53	64	69	72	75	...	38	36	33	32	39	45	68	60	45	83
6	173	30	37	41	42	57	81	88	77	64	...	31	32	35	32	35	49	65	64	53	87
7	171	24	32	36	42	55	77	101	107	102	...	54	64	63	51	53	60	56	46	49	89
8	170	21	31	29	28	35	47	62	76	83	...	105	101	89	63	45	42	41	37	61	101
9	172	21	22	27	28	30	33	43	46	44	...	129	118	103	74	39	27	36	34	68	101
10	171	23	30	21	30	36	44	51	46	41	...	136	126	120	102	70	40	30	35	78	101
11	175	21	33	31	42	54	71	73	64	59	...	142	133	122	114	111	77	31	36	95	104
12	192	42	27	37	63	87	99	111	116	116	...	105	108	108	107	109	104	58	59	100	97
13	196	82	41	58	102	112	110	110	107	108	...	70	71	86	102	107	116	84	77	95	99
14	192	88	78	95	96	87	81	72	51	40	...	62	73	73	101	116	116	99	80	88	105
15	190	88	102	114	99	76	55	55	50	37	...	100	112	120	120	125	125	105	97	101	98
16	189	106	111	113	137	124	113	109	103	96	...	134	137	142	146	134	120	96	115	122	93
17	188	106	132	119	142	160	158	153	148	145	...	163	158	156	142	124	110	100	114	108	86
18	193	83	140	130	122	141	153	160	168	177	...	163	158	148	129	112	103	99	99	78	77
19	190	81	117	127	107	120	134	146	163	166	...	157	145	132	117	105	103	90	64	67	72
20	193	83	84	106	104	113	122	134	138	143	...	142	129	116	109	105	102	86	55	60	63
21	194	78	87	91	92	108	113	122	127	140	...	140	128	118	113	109	101	68	56	56	56
22	191	80	89	88	90	104	114	120	131	141	...	130	129	119	108	103	101	50	53	55	53
23	189	77	89	91	86	93	111	122	133	129	...	102	113	111	107	101	85	53	51	54	55
24	190	86	88	87	87	87	104	115	127	115	...	105	108	104	102	97	55	53	50	53	59
25	187	74	89	81	93	130	103	96	110	108	...	111	105	103	102	83	50	49	56	51	49
26	190	79	81	107	166	206	119	88	94	105	...	102	104	99	100	111	111	57	48	52	53
27	192	78	83	173	211	158	114	100	87	94	...	101	98	98	96	116	123	119	52	49	55
28	188	70	136	177	198	108	101	119	86	81	...	98	98	93	82	80	123	145	73	43	51
29	196	87	193	187	179	113	123	123	110	81	...	95	90	96	77	53	160	124	103	66	52
30	219	179	202	196	198	146	122	118	119	94	...	92	90	87	57	89	126	140	150	127	109
31	244	228	230	231	233	213	188	195	193	189	...	179	184	177	161	202	182	207	198	202	206

32 rows × 32 columns

# 上面获得的32*32的二维数组，还不利于数据建模，所以我们还需要通过reshape(1, -1)方法将其转换成一行(若reshape(-1,1)则转为一列），也即1*1024格式
arr = arr.reshape(1, -1)print(arr)  # 查看转换后的结果，这一行数就是代表那张人脸图片了，其共有32*32=1024列数

[[186  76  73 ... 198 202 206]]

因为总共有400张照片需要处理，若将400个二维数组堆叠起来会形成三维数组，因为我们需要使用flatten()函数将1*1024的二维数组降维成一维数组，并通过tolist()函数将其转为列表方便之后和其他图片的颜色数值信息一起处理。

print(arr.flatten().tolist())  # 下面这一行数就是那张人脸转换后的结果了

[186, 76, 73, 87, 89, 88, 75, 81, 100, 102, 105, 92, 74, 65, 65, 53, 43, 55, 53, 42, 58, 77, 71, 75, 75, 73, 76, 85, 95, 100, 103, 106, 196, 85, 68, 78, 104, 97, 100, 94, 83, 87, 88, 89, 86, 70, 65, 61, 55, 52, 38, 32, 52, 66, 52, 59, 70, 85, 62, 82, 89, 85, 106, 103, 193, 69, 79, 92, 105, 102, 112, 117, 106, 94, 91, 112, 101, 87, 75, 61, 58, 54, 49, 48, 44, 41, 41, 45, 50, 76, 59, 74, 83, 82, 99, 100, 186, 67, 71, 75, 85, 99, 114, 115, 109, 109, 98, 101, 86, 68, 74, 65, 58, 53, 51, 52, 42, 40, 42, 43, 40, 52, 41, 61, 69, 76, 76, 108, 179, 46, 41, 50, 53, 69, 80, 91, 108, 104, 98, 93, 91, 88, 73, 60, 56, 55, 51, 49, 53, 55, 43, 37, 30, 31, 35, 43, 59, 61, 56, 101, 173, 33, 43, 49, 48, 53, 64, 69, 72, 75, 82, 84, 84, 82, 72, 75, 69, 71, 67, 56, 58, 55, 38, 36, 33, 32, 39, 45, 68, 60, 45, 83, 173, 30, 37, 41, 42, 57, 81, 88, 77, 64, 63, 64, 65, 64, 48, 48, 68, 62, 47, 50, 45, 31, 31, 32, 35, 32, 35, 49, 65, 64, 53, 87, 171, 24, 32, 36, 42, 55, 77, 101, 107, 102, 98, 83, 71, 64, 44, 48, 54, 64, 76, 65, 49, 48, 54, 64, 63, 51, 53, 60, 56, 46, 49, 89, 170, 21, 31, 29, 28, 35, 47, 62, 76, 83, 87, 78, 53, 58, 65, 83, 90, 97, 108, 101, 97, 105, 105, 101, 89, 63, 45, 42, 41, 37, 61, 101, 172, 21, 22, 27, 28, 30, 33, 43, 46, 44, 43, 46, 50, 63, 76, 87, 95, 104, 114, 120, 122, 125, 129, 118, 103, 74, 39, 27, 36, 34, 68, 101, 171, 23, 30, 21, 30, 36, 44, 51, 46, 41, 45, 52, 61, 70, 84, 101, 112, 119, 126, 131, 134, 138, 136, 126, 120, 102, 70, 40, 30, 35, 78, 101, 175, 21, 33, 31, 42, 54, 71, 73, 64, 59, 65, 78, 95, 106, 110, 116, 122, 130, 140, 143, 141, 142, 142, 133, 122, 114, 111, 77, 31, 36, 95, 104, 192, 42, 27, 37, 63, 87, 99, 111, 116, 116, 117, 117, 121, 122, 120, 119, 120, 121, 120, 113, 107, 106, 105, 108, 108, 107, 109, 104, 58, 59, 100, 97, 196, 82, 41, 58, 102, 112, 110, 110, 107, 108, 107, 107, 115, 126, 126, 118, 98, 87, 69, 57, 55, 55, 70, 71, 86, 102, 107, 116, 84, 77, 95, 99, 192, 88, 78, 95, 96, 87, 81, 72, 51, 40, 52, 53, 68, 89, 116, 111, 83, 61, 44, 55, 61, 42, 62, 73, 73, 101, 116, 116, 99, 80, 88, 105, 190, 88, 102, 114, 99, 76, 55, 55, 50, 37, 60, 53, 49, 84, 161, 155, 109, 94, 83, 76, 84, 88, 100, 112, 120, 120, 125, 125, 105, 97, 101, 98, 189, 106, 111, 113, 137, 124, 113, 109, 103, 96, 84, 84, 99, 123, 172, 176, 139, 130, 127, 120, 116, 115, 134, 137, 142, 146, 134, 120, 96, 115, 122, 93, 188, 106, 132, 119, 142, 160, 158, 153, 148, 145, 156, 151, 139, 139, 160, 177, 151, 138, 141, 153, 162, 164, 163, 158, 156, 142, 124, 110, 100, 114, 108, 86, 193, 83, 140, 130, 122, 141, 153, 160, 168, 177, 171, 154, 135, 137, 161, 176, 158, 146, 143, 142, 159, 164, 163, 158, 148, 129, 112, 103, 99, 99, 78, 77, 190, 81, 117, 127, 107, 120, 134, 146, 163, 166, 155, 134, 138, 137, 153, 164, 141, 132, 132, 127, 148, 156, 157, 145, 132, 117, 105, 103, 90, 64, 67, 72, 193, 83, 84, 106, 104, 113, 122, 134, 138, 143, 141, 138, 93, 69, 84, 86, 71, 62, 75, 106, 142, 146, 142, 129, 116, 109, 105, 102, 86, 55, 60, 63, 194, 78, 87, 91, 92, 108, 113, 122, 127, 140, 148, 154, 121, 80, 72, 66, 80, 96, 106, 112, 137, 147, 140, 128, 118, 113, 109, 101, 68, 56, 56, 56, 191, 80, 89, 88, 90, 104, 114, 120, 131, 141, 145, 150, 153, 142, 144, 124, 105, 111, 119, 121, 128, 128, 130, 129, 119, 108, 103, 101, 50, 53, 55, 53, 189, 77, 89, 91, 86, 93, 111, 122, 133, 129, 125, 126, 118, 117, 115, 111, 100, 85, 87, 86, 76, 88, 102, 113, 111, 107, 101, 85, 53, 51, 54, 55, 190, 86, 88, 87, 87, 87, 104, 115, 127, 115, 101, 88, 83, 84, 91, 96, 108, 114, 101, 98, 106, 103, 105, 108, 104, 102, 97, 55, 53, 50, 53, 59, 187, 74, 89, 81, 93, 130, 103, 96, 110, 108, 108, 129, 126, 112, 119, 108, 96, 98, 105, 110, 117, 118, 111, 105, 103, 102, 83, 50, 49, 56, 51, 49, 190, 79, 81, 107, 166, 206, 119, 88, 94, 105, 112, 116, 113, 106, 96, 92, 95, 102, 103, 107, 113, 108, 102, 104, 99, 100, 111, 111, 57, 48, 52, 53, 192, 78, 83, 173, 211, 158, 114, 100, 87, 94, 108, 114, 115, 119, 129, 146, 151, 144, 140, 133, 121, 112, 101, 98, 98, 96, 116, 123, 119, 52, 49, 55, 188, 70, 136, 177, 198, 108, 101, 119, 86, 81, 94, 105, 115, 123, 127, 128, 126, 122, 119, 109, 97, 97, 98, 98, 93, 82, 80, 123, 145, 73, 43, 51, 196, 87, 193, 187, 179, 113, 123, 123, 110, 81, 73, 81, 88, 96, 97, 95, 91, 91, 89, 86, 91, 99, 95, 90, 96, 77, 53, 160, 124, 103, 66, 52, 219, 179, 202, 196, 198, 146, 122, 118, 119, 94, 76, 73, 72, 74, 77, 73, 74, 79, 77, 83, 92, 94, 92, 90, 87, 57, 89, 126, 140, 150, 127, 109, 244, 228, 230, 231, 233, 213, 188, 195, 193, 189, 181, 173, 171, 171, 171, 168, 168, 171, 173, 178, 182, 179, 179, 184, 177, 161, 202, 182, 207, 198, 202, 206]

# 构造所有图片的特征变量
X = []  # 特征变量
for i in names:img = Image.open('olivettifaces\\' + i)img = img.convert('L')img = img.resize((32, 32))arr = np.array(img)X.append(arr.reshape(1, -1).flatten().tolist())

import pandas as pd
X = pd.DataFrame(X)X  # 查看400张图片转换后的结果

	0	1	2	3	4	5	6	7	8	9	...	1014	1015	1016	1017	1018	1019	1020	1021	1022	1023
0	186	76	73	87	89	88	75	81	100	102	...	179	184	177	161	202	182	207	198	202	206
1	196	90	97	98	98	87	101	89	65	73	...	181	167	190	188	203	209	205	198	190	190
2	193	89	97	99	75	74	83	64	77	86	...	178	178	156	185	195	201	206	201	189	190
3	192	84	93	89	97	89	66	60	60	57	...	173	151	199	189	203	200	196	186	182	184
4	194	72	49	45	56	37	44	62	71	71	...	192	194	192	176	174	224	200	218	176	168
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
395	114	115	115	119	115	120	117	118	113	112	...	190	193	169	141	142	144	143	141	143	215
396	115	118	117	117	116	118	117	119	117	116	...	187	189	183	216	189	193	148	144	142	212
397	113	116	113	117	114	121	121	120	121	114	...	184	188	185	221	203	192	144	143	137	212
398	110	109	109	110	110	112	112	113	113	111	...	172	171	209	212	175	136	142	141	137	213
399	105	107	111	112	113	113	113	116	116	107	...	181	184	220	188	140	139	142	141	138	213

400 rows × 1024 columns

print(X.shape)  # 查看此时的表格结构

(400, 1024)

3. 人脸数据处理: 目标变量提取

# 获取目标变量y：第一张图片演示
print(int(names[0].split('_')[0]))

# 批量获取所有图片的目标变量y
y = []  # 目标变量
for i in names:img = Image.open('olivettifaces\\' + i)y.append(int(i.split('_')[0]))print(y)  # 查看目标变量,也就是对应的人员编号

[10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 17, 17, 17, 17, 17, 17, 17, 17, 17, 17, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 26, 26, 26, 26, 26, 26, 26, 26, 26, 26, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 28, 28, 28, 28, 28, 28, 28, 28, 28, 28, 29, 29, 29, 29, 29, 29, 29, 29, 29, 29, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 31, 31, 31, 31, 31, 31, 31, 31, 31, 31, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 33, 33, 33, 33, 33, 33, 33, 33, 33, 33, 34, 34, 34, 34, 34, 34, 34, 34, 34, 34, 35, 35, 35, 35, 35, 35, 35, 35, 35, 35, 36, 36, 36, 36, 36, 36, 36, 36, 36, 36, 37, 37, 37, 37, 37, 37, 37, 37, 37, 37, 38, 38, 38, 38, 38, 38, 38, 38, 38, 38, 39, 39, 39, 39, 39, 39, 39, 39, 39, 39, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9]

13.2.3 数据划分与降维

1. 划分训练集和测试集

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=1)

2. PCA数据降维

# 数据降维模型训练
from sklearn.decomposition import PCA
pca = PCA(n_components=100)
pca.fit(X_train)

PCA(n_components=100)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

PCA

PCA(n_components=100)

# 对训练集和测试集进行数据降维
X_train_pca = pca.transform(X_train)
X_test_pca = pca.transform(X_test)

# 我们通过如下代码验证PCA是否降维：
print(X_train_pca.shape)
print(X_test_pca.shape)

(320, 100)
(80, 100)

# 如果想查看此时降维后的X_train_pca和X_test_pca，可以直接将它们打印出来查看，也可以将它们转为DataFrame格式进行查看，代码如下：
# pd.DataFrame(X_train_pca).head()
# pd.DataFrame(X_test_pca).head()

# 在PCA后面加个“？”运行可以可以看看官方的一些提示
# PCA?

13.2.4 模型的搭建与使用

1. 模型搭建

from sklearn.neighbors import KNeighborsClassifier
knn = KNeighborsClassifier()  # 建立KNN模型  
knn.fit(X_train_pca, y_train)  # 用降维后的训练集进行训练模型

KNeighborsClassifier()

KNeighborsClassifier

KNeighborsClassifier()

2. 模型预测

y_pred = knn.predict(X_test_pca)  # 用降维后的测试集进行测试
print(y_pred)  # 将对测试集的预测结果打印出来

[ 9 21  3 40 26  4 28 37 12 36 26  7 27 21  3 24  7  2 17 24 21 32  8  211 19  6 29  6 29 18 10 25 35 10 18 15  5  9 22 34 29  2 16  8 18  8 3839 35 16 30 30 11 37 36 35 20 33  6  1 16 31 32  5 30  1 39 35 39  2 195  8 11  4 14 27 22 24]

# 通过和之前章节类似的代码，我们可以将预测值和实际值进行对比：
import pandas as pd
a = pd.DataFrame()  # 创建一个空DataFrame 
a['预测值'] = list(y_pred)
a['实际值'] = list(y_test)a.head()  # 查看表格前5行

	预测值	实际值
0	9	9
1	21	21
2	3	3
3	40	40
4	26	26

# 查看预测准确度 - 方法1
from sklearn.metrics import accuracy_score
score = accuracy_score(y_pred, y_test)
print(score)

0.9125

# 查看预测准确度 - 方法2
score = knn.score(X_test_pca, y_test)
print(score)

0.9125

3. 模型对比（数据降维与不降维）

from sklearn.neighbors import KNeighborsClassifier
knn = KNeighborsClassifier()  # 建立KNN模型  
knn.fit(X_train, y_train)  # 不使用数据降维，直接训练
y_pred = knn.predict(X_test)  # 不使用数据降维，直接测试from sklearn.metrics import accuracy_score
score = accuracy_score(y_pred, y_test)
print(score)

0.9125

此时获得的准确度评分score为0.91，可以看到使用数据降维对提高模型预测效果还是有一些效果的，这里的数据量并不大，当数据量更大的时候，利用PCA主成分分析进行数据降维则会发挥更大的作用。

13.3 补充知识点：人脸识别外部接口调用

13.3.1 baidu-aip库安装

pip install baidu-aip

13.3.2 调用接口进行人脸识别和打分

from aip import AipFace
import base64# 下面3行内容为自己的APP_ID,API_KEY,SECRET_KEY
APP_ID = '16994639'    #  '25572406'
API_KEY = 'L9XnkKQEMnHhB5omF2P8D9OM'   #'FFg2mnhYI8dD46uenYXjzhlg'
SECRET_KEY = 'nnOZDoruZ6AjVglBs6ecvUjFRIAKrn9T' #'ra9rPr3E4wuonUAEB7X1x41aKnh7qFy7'# 把上面输入的账号信息传入接口
aipFace = AipFace(APP_ID, API_KEY, SECRET_KEY)# 下面一行内容为需要识别的人脸图片的地址，其他地方就不用改了
filePath = r'张起凡.jpg'# 定义打开文件的函数
def get_file_content(filePath):with open(filePath, 'rb') as fp:content = base64.b64encode(fp.read())return content.decode('utf-8')imageType = "BASE64"# 选择最后要展示的内容，这里展示age（年龄）；gender（性别）；beauty（颜值）
options = {}
options["face_field"] = "age,gender,beauty"# 调用接口aipFace的detect()函数进行人脸识别，打印结果
result = aipFace.detect(get_file_content(filePath), imageType, options)
print(result)# 打印具体信息，本质就是列表索引和字典的键值对应
age = result['result']['face_list'][0]['age']
print('年龄预测为：' + str(age))
gender = result['result']['face_list'][0]['gender']['type']
print('性别预测为：' + gender)
beauty = result['result']['face_list'][0]['beauty']
print('颜值评分为：' + str(beauty))

{'error_code': 0, 'error_msg': 'SUCCESS', 'log_id': 943952634, 'timestamp': 1682385343, 'cached': 0, 'result': {'face_num': 1, 'face_list': [{'face_token': '8898d344d778dd12bcbc4efe488da9aa', 'location': {'left': 313.02, 'top': 379.26, 'width': 328, 'height': 309, 'rotation': 14}, 'face_probability': 1, 'angle': {'yaw': -23.27, 'pitch': 7.04, 'roll': 13.28}, 'age': 22, 'gender': {'type': 'male', 'probability': 0.99}, 'beauty': 53.09}]}}
年龄预测为：22
性别预测为：male
颜值评分为：53.09