def ntrial(numTrial, numTrain, numTest):
for i in range(numTrial):
#data
[X, Y, T] = dt.sample_poly3(numTrain+numTest);
[xTrain, yTrain, tTrain, xTest, yTest, tTest] = util.sepTrainTest4Reg(numTrain,numTest,\
X,T,Y);
errTrainPoly, errTestPoly = pR.run (xTrain, yTrain, tTrain, xTest, yTest, tTest, 4);
#errTrainPoly, errTestPoly = ppR.run(xTrain, yTrain, tTrain, xTest, yTest, tTest, 4);
errTrainBay, errTestBay = bcf.run(xTrain, yTrain, tTrain, xTest, yTest, tTest, 4);
errPoly[0].append(round(errTrainPoly,6));
errPoly[1].append(round(errTestPoly,6));
errBay[0].append(round(errTrainBay,6));
errBay[1].append(round(errTestBay,6));
return errPoly, errBay;
t = pl.plot(xTest, yTest, 'go', label="test data")
o = pl.plot(xTest, oTest, 'ro', label="predicted")
#pl.legend((t,o), ("target", "predicted"));
pl.title('Polynomial Regression test set with degree ' + str(K - 1))
if __name__ == '__main__':
K = 4
kList = [3]
errorTrain = [0] * len(kList)
errorTest = [0] * len(kList)
#data
[X, Y, T] = dt.sample_poly3(60)
[xTrain, yTrain, tTrain, xTest, yTest, tTest] = util.sepTrainTest4Reg(40,20,\
X,T,Y)
oTrain, mean = bayesianCurveFitting(xTrain, np.array(yTrain))
#graph
pl.figure(1)
t = pl.plot(xTrain, yTrain, 'go')
o = pl.plot(xTrain, oTrain, 'ro-')
#pl.plot(xTrain, oTrain2, 'bo-');
#pl.legend((t, o), ("target", "predicted"));
pl.ylim([-8, 5])
pl.title('Bayesian Regression train set with K=' + str(K))
Basis = basisX(xTest)
N = Basis.shape[1]
oTest = eval1(np.array(yTest), Basis.T, mean, N)
# evaluating train
oTrain = NPolynomReg(K, xTrain, yTrain, beta, w)
print "Euclidean error distance for train: "
print np.linalg.norm(np.array(yTrain) - oTrain)
# evaluating test
oTest = NPolynomReg(K, xTest, yTest, beta, w)
print "Euclidean error distance for test: "
print np.linalg.norm(np.array(oTest) - yTest)
# graph
pl.figure(3)
t = pl.plot(xTrain, yTrain, "go", label="train data")
pl.plot(xTrain, oTrain2, "bo-", label="curve fit")
# pl.legend((t, o), ("target", "predicted"));
# pl.ylim([-8,5]);
pl.title("Polynomial Regression train set with degree " + str(K - 1))
pl.figure(4)
t = pl.plot(xTest, yTest, "go", label="test data")
o = pl.plot(xTest, oTest, "ro", label="predicted")
# pl.legend((t,o), ("target", "predicted"));
pl.title("Polynomial Regression test set with degree " + str(K - 1))
if __name__ == "__main__":
[X, Y, T] = dt.sample_poly3(100)
[xTrain, yTrain, tTrain, xTest, yTest, tTest] = util.sepTrainTest4Reg(50, 50, X, T, Y)
errTrainPoly, errTestPoly = ppR.run(xTrain, yTrain, tTrain, xTest, yTest, tTest, 4)
pl.show()
o = pl.plot(xTest, oTest, 'ro', label="predicted");
#pl.legend((t,o), ("target", "predicted"));
pl.title('Polynomial Regression test set with degree ' + str(K-1));
if __name__ == '__main__':
K = 4;
kList = [3];
errorTrain = [0]*len(kList);
errorTest = [0]*len(kList);
#data
[X, Y, T] = dt.sample_poly3(60);
[xTrain, yTrain, tTrain, xTest, yTest, tTest] = util.sepTrainTest4Reg(40,20,\
X,T,Y);
oTrain, mean = bayesianCurveFitting(xTrain, np.array(yTrain));
#graph
pl.figure(1);
t = pl.plot(xTrain, yTrain, 'go');
o = pl.plot(xTrain, oTrain, 'ro-');
#pl.plot(xTrain, oTrain2, 'bo-');
#pl.legend((t, o), ("target", "predicted"));
pl.ylim([-8,5]);
pl.title('Bayesian Regression train set with K=' + str(K));
Basis = basisX(xTest);
N = Basis.shape[1];
oTest = eval1(np.array(yTest), Basis.T, mean, N);