dt = ret.iloc[k * N:(k + 1) * N, :]
# demean to zero
dt = dt - dt.mean(axis=0)
samples[k, :, :] = dt
S[k, :, :] = np.cov(samples[k], rowvar=False, bias=True)
return S, samples, n
K = 13
N = 250
S, samples, n = filter_by_start(ret, start_date='2002-01-02', K=K, N=N)
Sinv = np.linalg.pinv(S, hermitian=True)
#%%
L1, L2, _ = lambda_grid(num1=7, num2=3, reg=reg)
grid1 = L1.shape[0]
grid2 = L2.shape[1]
AIC = np.zeros((grid1, grid2))
BIC = np.zeros((grid1, grid2))
Omega_0 = np.zeros((K, p, p))
Theta_0 = np.zeros((K, p, p))
for g1 in np.arange(grid1):
for g2 in np.arange(grid2):
lambda1 = L1[g1, g2]
lambda2 = L2[g1, g2]
sol, info = ADMM_MGL(S,
M = 10 reg = 'GGL' save = False Sigma, Theta = group_power_network(p, K, M) draw_group_heatmap(Theta, save=save) S, sample = sample_covariance_matrix(Sigma, N) S_train, sample_train = sample_covariance_matrix(Sigma, N_train) Sinv = np.linalg.pinv(S, hermitian=True) #%% # grid search for best lambda values with warm starts L1, L2, W2 = lambda_grid(num1=3, num2=9, reg=reg) grid1 = L1.shape[0] grid2 = L2.shape[1] ERR = np.zeros((grid1, grid2)) FPR = np.zeros((grid1, grid2)) TPR = np.zeros((grid1, grid2)) DFPR = np.zeros((grid1, grid2)) DTPR = np.zeros((grid1, grid2)) AIC = np.zeros((grid1, grid2)) BIC = np.zeros((grid1, grid2)) gammas = np.linspace(0, 1, 20) GAM = np.zeros((len(gammas), grid1, grid2)) Omega_0 = get_K_identity(K, p)
Sigma, Theta = time_varying_power_network(p, K, M)
#np.linalg.norm(np.eye(p) - Sigma@Theta)/np.linalg.norm(np.eye(p))
#single_heatmap_animation(Theta)
S, sample = sample_covariance_matrix(Sigma, N)
S_train, sample_train = sample_covariance_matrix(Sigma, N)
Sinv = np.linalg.pinv(S, hermitian = True)
results = {}
results['truth'] = {'Theta' : Theta}
#%%
# grid search for best lambda values with warm starts
L1, L2, _ = lambda_grid(num1 = 10, num2 = 5, reg = reg)
grid1 = L1.shape[0]; grid2 = L2.shape[1]
ERR = np.zeros((grid1, grid2))
FPR = np.zeros((grid1, grid2))
TPR = np.zeros((grid1, grid2))
DFPR = np.zeros((grid1, grid2))
DTPR = np.zeros((grid1, grid2))
AIC = np.zeros((grid1, grid2))
BIC = np.zeros((grid1, grid2))
Omega_0 = get_K_identity(K,p)
Theta_0 = get_K_identity(K,p)
for g1 in np.arange(grid1):
for g2 in np.arange(grid2):