sigma = ul.fnp(np.sqrt(np.diag(Cov)))
## Plot the results
gl.plot_timeRegression(x_grid,
y_pred,
sigma,
dates,
timeSeries,
sigma_eps,
labels=["GP own implementation", "Time", "Price"],
nf=1)
optflag = 1
if (optflag):
### Optimize the parameters and do it again ####
xopt = gp.optimize_parameters(sigma_0, l, sigma_eps)
sigma_0, l, sigma_eps = xopt
gp.fit(dates,
timeSeries,
kernel=GPown.compute_Kernel,
params=dict([["l", l], ["sigma_0", sigma_0]]),
sigma_eps=sigma_eps)
## Generate a Validation set and obtain the regressed values !
y_pred, Cov = gp.predict(x_grid)
sigma = ul.fnp(np.sqrt(np.diag(Cov)))
gl.plot_timeRegression(x_grid,
y_pred,
sigma,
dates,
timeSeries,
sigma_eps,
params = dict([["l",l],["sigma_0",sigma_0]]),
sigma_eps = sigma_eps)
y_pred, Cov = gp.predict(x_grid)
sigma = ul.fnp(np.sqrt(np.diag(Cov)))
## Plot the results
gl.plot_timeRegression(x_grid, y_pred, sigma,
dates, timeSeries,sigma_eps,
labels = ["GP own implementation","Time","Price"], nf = 1)
optflag = 1
if (optflag):
### Optimize the parameters and do it again ####
xopt = gp.optimize_parameters(sigma_0, l, sigma_eps)
sigma_0, l, sigma_eps = xopt
gp.fit(dates,timeSeries,
kernel = GPown.compute_Kernel,
params = dict([["l",l],["sigma_0",sigma_0]]),
sigma_eps = sigma_eps)
## Generate a Validation set and obtain the regressed values !
y_pred, Cov = gp.predict(x_grid)
sigma = ul.fnp(np.sqrt(np.diag(Cov)))
gl.plot_timeRegression(x_grid, y_pred, sigma,
dates, timeSeries,sigma_eps,
labels = ["GP Estimation (Opt)","Time","Price"], nf = 1)
## Plot realizations !!
# f_s = gp.generate_process(x_grid, N = 20)
# gl.plot(x_grid,f_s, legend = ["Realization"], nf = 0)
