Make prediction for funcge model

[Jax-Accpet1]

Hi GENetLib team, I've read your update about predictions on scalarge model, but I still wonder how to make predictions on funcge model.

[ruiqwy]

Thank you for your question! We have thought of this issue as well. Here is an example code:

from GENetLib.sim_data import sim_data_func
from GENetLib.grid_func_ge import grid_func_ge
from GENetLib.fda_func import dense_to_func, create_bspline_basis, inprod
from GENetLib.pre_data import pre_data2
num_hidden_layers = 2
nodes_hidden_layer = [100, 10]
learning_rate2 = [0.008, 0.009, 0.01]
Lambda = [0.002, 0.003, 0.004, 0.005, 0.006]
learning_rate1 = [0.02, 0.03, 0.04, 0.05]
lambda2 = [0.05, 0.06, 0.07, 0.08]
num_epochs = 100
nbasis1 = 7
params1 = 4
func_continuous = sim_data_func(n = 1000, m = 100, ytype = 'Continuous', seed = 1)
y = func_continuous['y']
Z = func_continuous['Z']
location = func_continuous['location']
X = func_continuous['X']
grid_func_ge_res = grid_func_ge(y, X, location, Z, 'Continuous', 'Bspline', 
                      num_hidden_layers, nodes_hidden_layer, num_epochs, learning_rate1, learning_rate2, 
                      nbasis1, params1, lambda1 = None, lambda2 = lambda2, Lambda = Lambda,
                      Bsplines = 15, norder1 = 4, model = None, split_type = 1, ratio = [3, 1, 1], plot_res = False, plot_beta = True)
funcX = dense_to_func(location, X, btype = "Bspline", nbasis = 7, params = 4, Plot = False)
fbasis1 = create_bspline_basis(rangeval=[min(location), max(location)], nbasis=nbasis1, norder=params1)
fbasis2 = create_bspline_basis(rangeval=[min(location), max(location)], nbasis=15, norder=4)
n,m = X.shape
funcCoef = funcX['coefs'].T
basisint = inprod(fdobj1=fbasis1, fdobj2=fbasis2, Lfdobj1=0, Lfdobj2=0)
def funcU(i):
    return np.dot(funcCoef[i, :], basisint)
U = pd.DataFrame(np.array([funcU(i) for i in range(n)]).reshape(n, -1))
GE = np.zeros(shape=(n, U.shape[1] * Z.shape[1]))
k = 0
for i in range(Z.shape[1]):
    for j in range(U.shape[1]):
        GE[:,k] = Z[:,i] * U.iloc[:,j]
        k = k + 1
U_, y_, Z_, GE_ = pre_data2(y, U, Z, GE, ytype = 'Continuous', split_type = 0, ratio = [10, 0])[:4]
pred = grid_func_ge_res[0][5](U_, GE_, Z_)
from sklearn.metrics import r2_score
r2 = r2_score(y_.detach().numpy(), pred.detach().numpy())
print(r2)

[ruiqwy]

We are considering add a prediction function for scalarge and funcge.