-
Notifications
You must be signed in to change notification settings - Fork 0
/
anova_fig1.py
196 lines (159 loc) · 7.49 KB
/
anova_fig1.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
import numpy as np
from statsmodels.formula.api import ols
from statsmodels.stats.anova import anova_lm
from statsmodels.stats.multicomp import MultiComparison
import pandas as pd
from statsmodels.stats.multicomp import pairwise_tukeyhsd
ACC_score_list = np.load('./new_figure/figure_1.npy')
def ANOVA(df, alpha=0.05):
"""
:param df: 传入的df是一个字典，key为水平,value为每一次的trial
:param alpha: 传入置信度
:return:
"""
df = pd.DataFrame(df)
df_melt = df.melt()
df_melt.columns = ['Treat', 'Value']
model = ols('Value~C(Treat)', data=df_melt).fit()
anova_table = anova_lm(model, typ=2)
print(anova_table)
mc = MultiComparison(df_melt['Value'], df_melt['Treat'])
tukey_result = mc.tukeyhsd(alpha=alpha)
print(tukey_result)
return tukey_result
def two_way_ANOVA(df, alpha=0.1):
"""
aim:研究没有交互作用的双因素方差分析
:param df: 传入的df是一个dataframe数组，横轴和纵轴分别是两个因素
:param alpha: 传入置信度
:return:
"""
df1 = df.stack().reset_index().rename(columns={0: 'value'})
model = ols('value~C(A) + C(B)', df1).fit()
print(anova_lm(model))
print(pairwise_tukeyhsd(df1['value'], df1['B'], alpha=alpha)) # 第一个必须是销量， 也就是我们的指标。这个很正常，两两比较的方法一般比较保守
# df = {'60':[90,92,88],'65':[97,93,92],'70':[96,96,93],'75':[84,83,88],'80':[84,86,82]}
# ANOVA(df)
hg = {'tip':[89.3, 78.0, 77.1, 61.8], 'TMP': [92.8, 77.9, 72.4, 75.8], 'trp': [81.5, 71.0, 68.3, 67.5], 'kp': [84.0, 74.2, 70.6, 61.3]}
hg = pd.DataFrame(hg)
hg.index = pd.Index(['0','1','2','3'], name = 'A')
hg.columns = pd.Index(['tip','tmp','trp','kp'],name = 'B')
two_way_ANOVA(hg)
print(1)
feature_num = 7
sub_num = 14
RA_flag =6
mean_score = []
#two_way_ANOVA分析变量模型和变量subject对最终分类精度的影响
if RA_flag ==0:
PPG = ACC_score_list[:, range(sub_num)]
PPG_R = ACC_score_list[:, range(sub_num, 2 * sub_num)]
PPG_IR = ACC_score_list[:, range(2 * sub_num, 3 * sub_num)]
PPG_G = ACC_score_list[:, range(3 * sub_num, 4 * sub_num)]
PPG_RIR = ACC_score_list[:, range(4 * sub_num, 5 * sub_num)]
PPG_RG = ACC_score_list[:, range(5 * sub_num, 6 * sub_num)]
PPG_IRG = ACC_score_list[:, range(6 * sub_num, 7 * sub_num)]
d_PPG = pd.DataFrame(PPG)
d_PPG.index = pd.Index(['rf','lda','knn','svm'], name = 'A')
d_PPG.columns = pd.Index(['s1','s2','s3','s4','s5','s6','s7','s8','s9','s10','s11','s12','s13','s14'],name = 'B')
two_way_ANOVA(d_PPG)
#测试不同的模型之间是否有显著差异ANOVA,以及方便得出平均值写论文
if RA_flag==1:
PPG = ACC_score_list[:, range(sub_num)]
PPG_R = ACC_score_list[:, range(sub_num, 2 * sub_num)]
PPG_IR = ACC_score_list[:, range(2 * sub_num, 3 * sub_num)]
PPG_G = ACC_score_list[:, range(3 * sub_num, 4 * sub_num)]
PPG_RIR = ACC_score_list[:, range(4 * sub_num, 5 * sub_num)]
PPG_RG = ACC_score_list[:, range(5 * sub_num, 6 * sub_num)]
PPG_IRG = ACC_score_list[:, range(6 * sub_num, 7 * sub_num)]
feature_set_cata = [PPG, PPG_R, PPG_IR, PPG_G, PPG_RIR, PPG_RG, PPG_IRG]
for feature in feature_set_cata:
mean_fea = np.mean(feature, axis=1)
mean_score.append(mean_fea)
mean_score = np.array(mean_score)
print(1)
PPG = PPG*100
other_model = []
# for i in range(3):
# other_model.extend(list(PPG[i,:]))
#df = {'rf':list(PPG[0,[10,11]]), 'lda':list(PPG[1,[10,11]]),'knn':list(PPG[2,[10,11]]),'svm':list(PPG[3,[10,11]]),}
df = {'rf':list(PPG[0,0:2]), 'lda':list(PPG[1,0:2]),'knn':list(PPG[2,0:2]),'svm':list(PPG[3,0:2]),}
df = {'knn': list(PPG[2, 0:2]), 'svm': list(PPG[3, 0:2]), }
#df = {'rf': list(PPG_R[0, :]), 'lda': list(PPG_R[1, :]), 'knn': list(PPG_R[2, :]), 'svm': list(PPG_R[3, :]), }
ANOVA(df, alpha=0.1)
print(1)
#验证信号的组合是否有用
if RA_flag == 2:
PPG = ACC_score_list[:, range(sub_num)].ravel()
PPG_R = ACC_score_list[:, range(sub_num, 2 * sub_num)].ravel()
PPG_IR = ACC_score_list[:, range(2 * sub_num, 3 * sub_num)].ravel()
PPG_G = ACC_score_list[:, range(3 * sub_num, 4 * sub_num)].ravel()
PPG_RIR = ACC_score_list[:, range(4 * sub_num, 5 * sub_num)].ravel()
PPG_RG = ACC_score_list[:, range(5 * sub_num, 6 * sub_num)].ravel()
PPG_IRG = ACC_score_list[:, range(6 * sub_num, 7 * sub_num)].ravel()
feature_set_cata = [PPG, PPG_R, PPG_IR, PPG_G, PPG_RIR, PPG_RG, PPG_IRG]
#df = {'PPG':list(PPG), 'PPG_R':list(PPG_R), 'PPG_IR':list(PPG_IR), 'PPG_G':list(PPG_G), 'PPG_RIR':list(PPG_RIR), 'PPG_RG':list(PPG_RG), 'PPG_IRG':list(PPG_IRG)}
df = {'PPG_RIR': list(PPG_RIR), 'PPG_RG': list(PPG_RG), 'PPG_IRG': list(PPG_IRG)}
#df = {'PPG_R':list(PPG_R), 'PPG_IR':list(PPG_IR), 'PPG_G':list(PPG_G)}
ANOVA(df, alpha=0.05)
print(1)
if RA_flag == 3:
#研究结果2：手势分类分类器的影响，
sub_num = 14
rf_score = np.load('./new_figure/sub_score_array_rf.npy')
lda_score = np.load('./new_figure/sub_score_array_lda.npy')
knn_score = np.load('./new_figure/sub_score_array_knn.npy')
svm_score = np.load('./new_figure/sub_score_array_svm.npy')
model_score_list = [rf_score, lda_score, knn_score, svm_score]
gesture_acc_list = []
for i in range(4):
gesture_acc_list.append(model_score_list[i][:,0,0])
gesture_acc_list = np.array(gesture_acc_list)
d_PPG = pd.DataFrame(gesture_acc_list)
d_PPG.index = pd.Index(['rf', 'lda', 'knn', 'svm'], name='A')
d_PPG.columns = pd.Index(['s1', 's2', 's3', 's4', 's5', 's6', 's7', 's8', 's9', 's10', 's11', 's12', 's13', 's14'],
name='B')
two_way_ANOVA(d_PPG)
if RA_flag==4:
#研究结果3：研究力分类不同信号的影响
sub_num =14
ACC_overfeature_list = np.load('./new_figure/ACC_overfeature_list.npy')
d_PPG = pd.DataFrame(ACC_overfeature_list)
d_PPG.index = pd.Index(['PPG', 'R', 'IR', 'G'], name='A')
d_PPG.columns = pd.Index(['s1', 's2', 's3', 's4', 's5', 's6', 's7', 's8', 's9', 's10', 's11', 's12', 's13', 's14'],
name='B')
two_way_ANOVA(d_PPG)
if RA_flag==5:
#研究结果3：研究手势类型对力分类的影响
sub_num =14
ACC_score_list = np.load('./new_figure/Force_score_list.npy')
Force_score_list = []
mean_force_arr = np.zeros((16,4))
std_force_arr = np.zeros((16,4))
for r in range(int(ACC_score_list.shape[0]/sub_num)):
for c in range(4):
data = ACC_score_list[range(r*sub_num, r*sub_num+sub_num),c]
mean_force = np.mean(data)
std_force = np.std(data,ddof=1)
mean_force_arr[r,c] = mean_force
std_force_arr[r,c] = std_force
SVM_score = mean_force_arr[12:,:]
d_PPG = pd.DataFrame(SVM_score)
d_PPG.index = pd.Index(['PPG', 'R', 'IR', 'G'], name='A')
d_PPG.columns = pd.Index(['TIP', 'TMP', 'TRP', 'KP'],
name='B')
two_way_ANOVA(d_PPG)
if RA_flag==6:
#研究结果3：研究手势类型对力分类的影响，双因素分析，人+手势类型
sub_num =14
ACC_score_list = np.load('./new_figure/Force_score_list.npy')
Force_score_list = []
mean_force_arr = np.zeros((16,4))
std_force_arr = np.zeros((16,4))
ppg_all = ACC_score_list[168: 168+14,:]
d_PPG = pd.DataFrame(ppg_all)
d_PPG.index = pd.Index(list(np.arange(14)), name='A')
d_PPG.columns = pd.Index(['TIP', 'TMP', 'TRP', 'KP'],
name='B')
two_way_ANOVA(d_PPG)
print(1)