def traditional_surv_analysis(datas, opts):
# tidy data as ndarray
train_X, train_Y = datas["train"].xs.numpy(), datas["train"].ys.numpy()
test_X, test_Y = datas["test"].xs.numpy(), datas["test"].ys.numpy()
if "val" in datas.keys():
train_X = np.concatenate([train_X, datas["val"].xs])
train_Y = np.concatenate([train_Y, datas["val"].ys])
# construct structured array
train_Y = Surv.from_arrays(train_Y[:, 1].astype("bool"), train_Y[:, 0])
test_Y = Surv.from_arrays(test_Y[:, 1].astype("bool"), test_Y[:, 0])
# construct estimators
estimators = {
"CoxPH": CoxPHSurvivalAnalysis(),
"CGBSA": CGBSA(n_estimators=500, random_state=opts.random_seed),
"GBSA": GBSA(n_estimators=500, random_state=opts.random_seed),
"FKSVM": FKSVM(random_state=opts.random_seed),
"FSVM": FSVM(random_state=opts.random_seed)
}
# training
for name, estimator in estimators.items():
print("%s training." % name)
estimator.fit(train_X, train_Y)
# evaluation
train_scores = {}
test_scores = {}
for name, estimator in estimators.items():
print("%s evaluation." % name)
train_scores[name] = estimator.score(train_X, train_Y)
test_scores[name] = estimator.score(test_X, test_Y)
# return
return train_scores, test_scores
verbose=args.verbose,
minibatch_frac=1.0,
Base=base_name_to_learner[args.base],
Score=eval(args.score),
)
train_losses = ngb.fit(X_train, Y_train, E_train)
forecast = ngb.pred_dist(X_test)
train_forecast = ngb.pred_dist(X_train)
print("NGB score: %.4f (val), %.4f (train)" % (
concordance_index_censored(E_test.astype(bool), Y_test,
-forecast.mean())[0],
concordance_index_censored(E_train.astype(bool), Y_train,
-train_forecast.mean())[0],
))
##
## sksurv
##
gbsa = GBSA(
n_estimators=args.n_est,
learning_rate=args.lr,
subsample=args.minibatch_frac,
verbose=args.verbose,
)
gbsa.fit(X_train, Y_join(Y_train, E_train))
print("GBSA score: %.4f (val), %.4f (train)" % (
gbsa.score(X_test, Y_join(Y_test, E_test)),
gbsa.score(X_train, Y_join(Y_train, E_train)),
))