def main(): g = gd() patientv = [pinfo[1:] for pname, pinfo in g.REMISSED_PATIENTS.items()] print len(patientv[0]) m = model(len(patientv[0])-2) train_in = [x[:265] for x in patientv[:120]] train_out = [x[266] for x in patientv[:120]] train_set= zip(train_out,train_in) m.train(train_set,0.01) test = [x[:266] for x in patientv[120:]] for x in test: x = np.insert(x,0,1) print m.reg(x)
def main():
from read import getData as gd
g = gd()
tset = g.get_trainset(0.8, 265, vectorize=True)
train_data = tset[0]
test_data = tset[1]
n = Network([265, 100, 100, 2])
# train
n.train(train_data, 2000, 20, 0.0001)
for i in xrange(len(test_data)):
print n.feed(test_data[i][1])
print test_data[i][0]
def cost_derivative(self,x,y):
return (y - self.reg(x))
def cost_func(self,train_set):
suma = []
for y,x in train_set:
x = np.insert(x,0,1)
a = 1.0/2.0 *(y-self.reg(x))**2
suma.append(a)
sum = np.sum(suma)
return 1.0/len(train_set) * sum
def reg(self,x):
output = np.dot(self.parameters.transpose(),x)
#print output
return output
from read import getData as gd
if __name__ =='__main__':
g = gd(scale=True)
train_data = g.get_trainset(0.8,266)
train_set = train_data[0]
test_set = train_data[1]
n = RegModel(266)
y = test_set[0][1]
y = np.insert(y,0,1)
n.train(train_set,0.000001,1000)
x = test_set[0][0]
print n.reg(y)
print x
with open("parameters.txt",'w') as f:
f.write(np.array_str(n.parameters))