"""Generate submissions with DeepWalk + Node2Vec score features.""" from __future__ import annotations import argparse import importlib.util from pathlib import Path import lightgbm as lgb import numpy as np import pandas as pd from gensim.models import Word2Vec def load_module(name: str, path: Path): spec = importlib.util.spec_from_file_location(name, path) module = importlib.util.module_from_spec(spec) assert spec.loader is not None spec.loader.exec_module(module) return module def make_subs(root: Path, out_dir: Path, score: np.ndarray, ratios: list[float], thresholds: list[float]) -> None: known = np.load(root / "cached_scores" / "test_known_mask.npy").astype(bool) for ratio in ratios: pred = np.zeros(len(score), dtype=np.int8) pred[np.argsort(score)[-int(round(len(score) * ratio)):]] = 1 pred[known] = 1 path = out_dir / f"submission_content_mf_deepwalk_node2vec_lgb_r{ratio:.3f}.csv" pd.DataFrame({"Index": np.arange(len(pred), dtype=np.int64), "Predicted": pred}).to_csv(path, index=False) print(path, int(pred.sum()), float(pred.mean())) for th in thresholds: pred = (score >= th).astype(np.int8) pred[known] = 1 path = out_dir / f"submission_content_mf_deepwalk_node2vec_lgb_th{th:.6f}.csv" pd.DataFrame({"Index": np.arange(len(pred), dtype=np.int64), "Predicted": pred}).to_csv(path, index=False) print(path, int(pred.sum()), float(pred.mean())) def main() -> None: parser = argparse.ArgumentParser() parser.add_argument("--package-root", type=Path, default=Path(__file__).resolve().parents[1]) parser.add_argument("--split-seed", type=int, default=202) parser.add_argument("--main-val-score-file", type=Path, required=True) parser.add_argument("--seed", type=int, default=202) parser.add_argument("--ratios", nargs="*", type=float, default=[0.498, 0.499, 0.500, 0.501, 0.502]) parser.add_argument("--thresholds", nargs="*", type=float, default=[0.455694, 0.48, 0.49, 0.50]) args = parser.parse_args() root = args.package_root stack = load_module("stack", root / "code" / "stack_rank_calibration.py") lgcn = load_module("lgcn", root / "code" / "train_val_lgcn_ensemble.py") post = load_module("post", root / "code" / "post95_ablation.py") gen = load_module("gen", root / "code" / "generate_post95_submission.py") extra = load_module("extra", root / "code" / "extra_score_sources_ablation.py") n2vmod = load_module("n2vmod", root / "code" / "node2vec_deepwalk_ablation.py") base_dir = root / "validation_runs" / f"dynamic_seed{args.split_seed}" / "node2vec_deepwalk" out_dir = root / "validation_runs" / f"dynamic_seed{args.split_seed}" / "node2vec_deepwalk_submission" out_dir.mkdir(parents=True, exist_ok=True) train_refs, val_pairs = lgcn.make_notebook_style_split(root, args.split_seed, 0.9) val_pairs_arr = val_pairs[["source", "target"]].to_numpy(np.int64) y = val_pairs["label"].to_numpy(np.int8) main_val = np.load(args.main_val_score_file).astype(np.float32) val_builder = stack.ExplicitGraphFeatures(root, train_refs) Xh = val_builder.transform(val_pairs_arr) X_val = np.column_stack( [ stack.add_rank_features(val_pairs_arr, main_val), Xh, post.negative_evidence_features(Xh, main_val), gen.topk_content_similarity_fast(root, val_pairs_arr, val_builder), ] ).astype(np.float32) selected = [Path(x.strip()) for x in (root / "validation_runs" / f"dynamic_seed{args.split_seed}" / "post95_submission" / "selected_variant_val_scores.txt").read_text().splitlines() if x.strip()] X_val = np.column_stack([X_val, gen.variant_feature_matrix(post, [np.load(p).astype(np.float32) for p in selected])]).astype(np.float32) content_val = extra.content_mean_score(root, val_pairs_arr, val_builder) mf_val = np.load(root / "validation_runs" / f"dynamic_seed{args.split_seed}" / "extra_score_sources" / f"val_mf_bpr_s{args.seed}_d256.npy").astype(np.float32) Xc, _ = n2vmod.score_to_features(content_val, "content_mean_cos", val_pairs_arr) Xm, _ = n2vmod.score_to_features(mf_val, "mf_bpr", val_pairs_arr) X_val = np.column_stack([X_val, Xc, Xm]).astype(np.float32) for name in ["deepwalk", "node2vec"]: cos = np.load(base_dir / f"{name}_cos_{len(val_pairs_arr)}_{int(val_pairs_arr[:,0].sum())}_{int(val_pairs_arr[:,1].sum())}.npy") dot = np.load(base_dir / f"{name}_dot_{len(val_pairs_arr)}_{int(val_pairs_arr[:,0].sum())}_{int(val_pairs_arr[:,1].sum())}.npy") Xcos, _ = n2vmod.score_to_features(cos, f"{name}_cos", val_pairs_arr) Xdot, _ = n2vmod.score_to_features(dot, f"{name}_dot", val_pairs_arr) X_val = np.column_stack([X_val, Xcos, Xdot]).astype(np.float32) print("fit", X_val.shape) clf = lgb.LGBMClassifier( n_estimators=1200, learning_rate=0.025, num_leaves=31, subsample=0.9, colsample_bytree=0.9, reg_lambda=5.0, min_child_samples=80, objective="binary", verbose=-1, random_state=args.seed, ) clf.fit(X_val, y) test_pairs = np.array(gen.read_txt(root / "data_and_docs" / "bipartite_test_ann.txt"), dtype=np.int64) main_test = np.load(root / "validation_runs" / f"dynamic_seed{args.split_seed}" / "post95_test_scores" / "dyn202_l2d512_bpr_bigbatch_more" / "scores" / "test_vanilla_ensemble_mean.npy").astype(np.float32) full_refs = pd.DataFrame(gen.read_txt(root / "data_and_docs" / "bipartite_train_ann.txt"), columns=["source", "target"]) test_builder = stack.ExplicitGraphFeatures(root, full_refs) Xht = test_builder.transform(test_pairs) X_test = np.column_stack( [ stack.add_rank_features(test_pairs, main_test), Xht, post.negative_evidence_features(Xht, main_test), gen.topk_content_similarity_fast(root, test_pairs, test_builder), ] ).astype(np.float32) test_scores = [] for p in selected: rel = p.resolve().relative_to(root / "validation_runs" / f"dynamic_seed{args.split_seed}") tp = root / "validation_runs" / f"dynamic_seed{args.split_seed}" / "post95_test_scores" / rel.parent / rel.name.replace("val_", "test_", 1) test_scores.append(np.load(tp).astype(np.float32)) X_test = np.column_stack([X_test, gen.variant_feature_matrix(post, test_scores)]).astype(np.float32) content_test = extra.content_mean_score(root, test_pairs, test_builder) mf_test = np.load(root / "validation_runs" / f"dynamic_seed{args.split_seed}" / "extra_bprmf_submission" / "test_mf_bpr_dynamic_s202_d256_e220.npy").astype(np.float32) Xct, _ = n2vmod.score_to_features(content_test, "content_mean_cos", test_pairs) Xmt, _ = n2vmod.score_to_features(mf_test, "mf_bpr", test_pairs) X_test = np.column_stack([X_test, Xct, Xmt]).astype(np.float32) for name, model_file in [ ("deepwalk", base_dir / "deepwalk_d128.model"), ("node2vec", base_dir / "node2vec_d128_p1_q2.model"), ]: model = Word2Vec.load(str(model_file)) cos, dot = n2vmod.pair_scores(model, test_pairs, name, root, args.split_seed) Xcos, _ = n2vmod.score_to_features(cos, f"{name}_cos", test_pairs) Xdot, _ = n2vmod.score_to_features(dot, f"{name}_dot", test_pairs) X_test = np.column_stack([X_test, Xcos, Xdot]).astype(np.float32) print("predict", X_test.shape) score = clf.predict_proba(X_test)[:, 1].astype(np.float32) np.save(out_dir / "test_content_mf_deepwalk_node2vec_lgb_pred.npy", score) make_subs(root, out_dir, score, args.ratios, args.thresholds) if __name__ == "__main__": main()