def main():
(x, y) = LSM.data()
(xwn, ywn) = LSM.data(Noise=True)
lsm = LSM.LSM_L2(6)
lsm.fit(x, y)
lsm2 = LSM.LSM_L2(6)
lsm2.fit(xwn, ywn)
tics = 2 * pi * arange(0, 1, 0.02)
lw = 5 #line width
plt.subplot(211)
plt.legend()
plt.plot(x, y, 'ro', ms=10)
plt.plot(tics, sin(tics), linewidth=lw, label="sin")
plt.plot(tics, lsm.predict(tics), linewidth=lw, label="fitting")
plt.legend(loc='lower right')
plt.subplot(212)
plt.xlim(0, 2 * pi)
plt.ylim(-2, 2)
plt.legend()
plt.plot(xwn, ywn, 'ro', ms=10)
plt.plot(tics, sin(tics), linewidth=lw, label="sin")
plt.plot(tics, lsm2.predict(tics), linewidth=lw, label="fitting")
plt.legend(loc='lower right')
plt.show()
def main():
(x,y) = LSM.data()
(xwn, ywn) = LSM.data(Noise=True)
lsm = LSM.LSM_L2(6)
lsm.fit(x,y)
lsm2 = LSM.LSM_L2(6)
lsm2.fit(xwn,ywn)
tics = 2*pi*arange(0,1,0.02);
lw = 5 #line width
plt.subplot(211)
plt.legend()
plt.plot(x,y, 'ro', ms=10)
plt.plot(tics,sin(tics), linewidth=lw, label="sin")
plt.plot(tics,lsm.predict(tics), linewidth=lw, label="fitting");
plt.legend(loc = 'lower right')
plt.subplot(212)
plt.xlim(0,2*pi)
plt.ylim(-2,2)
plt.legend()
plt.plot(xwn,ywn, 'ro', ms=10)
plt.plot(tics,sin(tics), linewidth=lw, label="sin")
plt.plot(tics,lsm2.predict(tics), linewidth=lw, label="fitting");
plt.legend(loc = 'lower right')
plt.show()
def main():
(x,y) = LSM.data()
(xwn, ywn) = LSM.data(Noise=True)
lsm = LSM.LSM_L2(M=7, l=2.)
tics = 2*pi*arange(0,1,0.02);
lw = 5 #line width
ls = power(7., range(-2,3))
print(ls)
for idcs in range(len(ls)):
plt.subplot(len(ls),1,idcs+1)
plt.xlim(0,2*pi+2)
plt.ylim(-2,2)
plt.legend()
plt.plot(xwn,ywn, 'ro', ms=10)
plt.plot(tics,sin(tics), linewidth=lw, label="sin")
lsm.l = ls[idcs]
lsm.fit(xwn,ywn)
plt.plot(tics,lsm.predict(tics), linewidth=lw, label='l={:.1}'.format(ls[idcs]));
plt.legend(loc = 'lower right')
plt.show()
def main():
(x, y) = LSM.data()
(xwn, ywn) = LSM.data(Noise=True, NL=0.3, N=10)
lsm = LSM.LSM_L2(M=7, l=2.)
tics = 2 * pi * arange(0, 1, 0.02)
lw = 5 #line width
ls = array([0.0, 0.1, 10, 100])
print(ls)
for idcs in range(len(ls)):
plt.subplot(len(ls), 1, idcs + 1)
plt.xlim(0, 2 * pi + 2)
plt.ylim(-2, 2)
plt.legend()
plt.plot(xwn, ywn, 'ro', ms=10)
plt.plot(tics, sin(tics), linewidth=lw, label="sin")
lsm.l = ls[idcs]
lsm.fit(xwn, ywn)
plt.plot(tics,
lsm.predict(tics),
linewidth=lw,
label='l={:.1}'.format(ls[idcs]))
plt.legend(loc='lower right')
plt.show()
# データをプロット
plt.plot(data_x, data_y, "o", label="data")
# 次数Nを決定
N = 3
# 次数0からNまで実行
for n in range(N+1) :
# 実行する次数
print("N="+str(n))
# LSMで誤差と重みベクトルcoeを計算
error, coe = LSM.LSM(data, int(n+1))
print("error:"+str(error))
print("coe:"+str(coe[::-1]))
# ここで，このモデルのAICを計算
l = AIC.l_MAX(list(coe[::-1]),data_x,data_y)
AIC_n = AIC.AIC(l,n+1)
print("l="+str(l)+", AIC("+str(n)+")="+str(AIC_n))
# LSMで得た近似線をプロット
test_x = np.arange(min(data_x), max(data_x), 0.01)
plt.plot(test_x, LSM.quation_LSM(coe, test_x), label="polynomial of degree "+str(n)+", AIC("+str(n)+")="+str(AIC_n))
plt.title("Compare LSM of degree")
plt.legend()
plt.grid()
# print(data)
# データをわかりやすい配列に格納(pltに使いやすくしたいから)
data_x = []
data_y = []
for i in data:
data_x.append(i[0])
data_y.append(i[1])
# 次数Nを決定
N = 3
print("N=" + str(N))
# LSMで誤差と重みベクトルcoeを計算
error, coe = LSM.LSM(data, int(N + 1))
print("error:" + str(error))
print("coe:" + str(coe[::-1]))
# ここで，このモデルのAICを計算
l = AIC.l_MAX(list(coe[::-1]), data_x, data_y)
print("l=" + str(l) + ",AIC(" + str(N) + ")=" + str(AIC.AIC(l, N + 1)))
# データをプロット
plt.plot(data_x, data_y, "o")
# LSMで得た近似線をプロット
test_x = np.arange(min(data_x), max(data_x), 0.01)
plt.plot(test_x, LSM.quation_LSM(coe, test_x))
plt.grid()
data_y.append(i[1])
# データをプロット
plt.plot(data_x, data_y, "o", label="data")
# 次数Nを決定
N = 6
# 次数0からNまで実行
for n in range(N + 1):
# 実行する次数
print("N=" + str(n))
# LSMで誤差と重みベクトルcoeを計算
error, coe = LSM.LSM(data, int(n + 1))
print("error:" + str(error))
print("coe:" + str(coe[::-1]))
# ここで，このモデルのAICを計算
l = AIC.l_MAX(list(coe[::-1]), data_x, data_y)
AIC_n = AIC.AIC(l, n + 1)
print("l=" + str(l) + ", AIC(" + str(n) + ")=" + str(AIC_n))
# LSMで得た近似線をプロット
test_x = np.arange(min(data_x), max(data_x), 0.01)
plt.plot(test_x,
LSM.quation_LSM(coe, test_x),
label="polynomial of degree " + str(n) + ", AIC(" + str(n) +
")=" + str(AIC_n))