# simple basis functions
fns = [lambda j, x: x, lambda j, x: x**2, lambda j, x: x**3, lambda j, x: x**4]
if __name__ == '__main__':
np.random.seed(7)
# 100 data point
N = 100
M = len(fns)
s = 10
# the data to learn
x = s * np.random.rand(N)
w = np.random.rand()
y = np.sin(x) + 0.1 * np.random.randn(N)
# plot the raw data (sin + noise)
pl.plot(x, y, '.r')
# predict with regression
x0 = np.linspace(min(x), max(x), 500)
m = Regression(0.5, N, M, fns)
m.fit(x, y)
y0 = m.predict(x0)
pl.plot(x0, y0)
pl.show()
# Now set the values in the array:
ys[ln] = y
xs[ln] = x
ln += 1
return ys, xs
if __name__ == '__main__':
ys, xs = read_data()
N = 4000
M = 8 # from sex to shell height
# simple basis functions
fns = [lambda j, x: x[j]] * M
# train
model = Regression(5, N, M, fns)
model.fit(xs[0:N], ys[0:N])
# predict
predict_ys = model.predict(xs[N:])
# plot the result
x0 = range(0, len(predict_ys))
pl.plot(x0, ys[N:], '-r')
pl.plot(x0, predict_ys, '-')
pl.show()
xs[ln] = x
ln += 1
return ys, xs
if __name__ == '__main__':
ys, xs = read_data()
N = 4000
M = 8 # from sex to shell height
# simple basis functions
fns = [lambda j, x : x[j]] * M
# train
model = Regression(5, N, M, fns)
model.fit(xs[0:N], ys[0:N])
# predict
predict_ys = model.predict(xs[N:])
# plot the result
x0 = range(0, len(predict_ys))
pl.plot(x0, ys[N:], '-r')
pl.plot(x0, predict_ys, '-')
pl.show()
# simple basis functions fns = [lambda j, x : x, lambda j, x : x**2, lambda j, x: x**3, lambda j, x: x**4] if __name__ == '__main__': np.random.seed(7) # 100 data point N = 100 M = len(fns) s = 10 # the data to learn x = s * np.random.rand(N) w = np.random.rand() y = np.sin(x) + 0.1 * np.random.randn(N) # plot the raw data (sin + noise) pl.plot(x,y,'.r') # predict with regression x0 = np.linspace(min(x),max(x),500) m = Regression(0.5, N, M, fns) m.fit(x, y) y0 = m.predict(x0) pl.plot(x0, y0) pl.show()
