def getAccs():
temp = []
for i in range(12):
t = []
for j in range(32):
u = []
for j in range(6):
u.append([])
t.append(u)
temp.append(t)
# get testaccs
for featIndex, featSet in enumerate(['ALL','EEG','PHY']):
for dimIndex, dim in enumerate([Classificators.ContValenceClassificator(), Classificators.ContArousalClassificator()]):
model = PersScript(featSet, 32, 30, dim, 1, "D:/ir-thesis/dumpedData/persScript/")
model.run()
for person, data in enumerate(model.accs):
for modelIndex, modelData in enumerate(data[0]):
temp[modelIndex][person][3*dimIndex+featIndex] = modelData[7]
names = np.array(['Pearson R', 'Mutual Information', 'distance Correlation', 'Linear Regression', 'Lasso regression',
'Ridgde regression', 'SVM', 'Random Forest', 'STD',
'ANOVA', 'LDA', 'PCA'])
f = open("temp.csv", 'w')
for mindex, model in enumerate(temp):
f.write('method;' + str(names[mindex]) + ';\n')
f.write(';valence;;;arousal;;;\n')
f.write('person;all;EEG;non-EEG;all;EEG;non-EEG;\n')
for pindex, person in enumerate(model):
f.write(str(pindex) + ';')
for values in person:
f.write(str(values) + ';')
f.write('\n')
f.write('\n\n\n')
f.close()
def phyeegall():
f = open("temp.csv", 'w')
# get testaccs
test_accs = []
all = []
model = PersScript("ALL", 32, 30, Classificators.ContValenceClassificator(),1, "D:/ir-thesis/dumpedData/persScript/")
model.run()
for person in model.accs:
all.append(person[0][7][7])
test_accs.append(all)
all = []
model = PersScript("EEG", 32, 30, Classificators.ContValenceClassificator(),1, "D:/ir-thesis/dumpedData/persScript/")
model.run()
for person in model.accs:
all.append(person[0][7][7])
test_accs.append(all)
all = []
model = PersScript("PHY", 32, 30, Classificators.ContValenceClassificator(),1, "D:/ir-thesis/dumpedData/persScript/")
model.run()
for person in model.accs:
all.append(person[0][7][7])
test_accs.append(all)
test_accs = np.array(test_accs)
avgs = np.average(test_accs, axis=1)
stds = np.std(test_accs, axis=1)
lbls = ['ALL','EEG','non-EEG']
genPlot(avgs,
stds,
lbls,
'Valence RF acc for different feat sets',
'feature Set',
'test acc',
['b','r','g']
)
print(avgs)
print(stds)
test_accs = []
all = []
model = PersScript("ALL", 32, 30, Classificators.ContArousalClassificator(),1,"D:/ir-thesis/dumpedData/persScript/")
model.run()
for person in model.accs:
all.append(person[0][7][7])
test_accs.append(all)
all = []
model = PersScript("EEG", 32, 30, Classificators.ContArousalClassificator(),1, "D:/ir-thesis/dumpedData/persScript/")
model.run()
for person in model.accs:
all.append(person[0][7][7])
test_accs.append(all)
all = []
model = PersScript("PHY", 32, 30, Classificators.ContArousalClassificator(),1, "D:/ir-thesis/dumpedData/persScript/")
model.run()
for person in model.accs:
all.append(person[0][7][7])
test_accs.append(all)
test_accs = np.array(test_accs)
avgs = np.average(test_accs, axis=1)
stds = np.std(test_accs, axis=1)
lbls = ['ALL', 'EEG', 'non-EEG']
genPlot(avgs,
stds,
lbls,
'Arousal RF acc for different feat sets',
'feature Set',
'test acc',
['b', 'r', 'g']
)
print(avgs)
print(stds)
f.close()
def svm_rbf_accs():
#get testaccs
test_accs = []
model = PersScript("ALL", 32, 30, Classificators.ContValenceClassificator(), "../dumpedData/persScript/")
model.run()
for person in model.accs:
model = person[0]
m = []
for metric in model:
m.append(metric[7])
test_accs.append(m)
test_accs = np.array(test_accs)
#test_accs = test_accs[:,np.array([0,1,2,3,4,5,6,7,9,10,11])]
sort_indices = np.array([0,1,2,9, 3,6,10, 4,5,7,11])
colors = ['b','b','b','b', 'r','r','r', 'g','g','g','g']
names = np.array(['R', 'MI', 'dC', 'LR', 'L1', 'L2', 'SVM', 'RF', 'STD', 'ANOVA', 'LDA', 'PCA'])
names = names[sort_indices]
test_accs = test_accs[:, sort_indices]
avgs = np.average(test_accs, axis=0)
stds = np.std(test_accs, axis=0)
genPlot(avgs,
stds,
names,
'Accuracies of valence SVM models',
'model',
'test acc',
colors
)
# get testaccs
test_accs = []
model = PersScript("ALL", 32, 30, Classificators.ContArousalClassificator(), "../dumpedData/persScript/")
model.run()
for person in model.accs:
model = person[0]
m = []
for metric in model:
m.append(metric[7])
test_accs.append(m)
test_accs = np.array(test_accs)
test_accs = test_accs[:, sort_indices]
#test_accs = test_accs[:, np.array([0, 1, 2, 3, 4, 5, 6, 7, 9, 10, 11])]
avgs = np.average(test_accs, axis=0)
stds = np.std(test_accs, axis=0)
genPlot(avgs,
stds,
names,
'Accuracy of arousal SVM models',
'model',
'test acc',
colors
)
print(names)
