| """Train/test split (by numeric case id range) + k-fold cross-validation.""" |
| import os, glob |
| import numpy as np |
| from .utils import numeric_id |
|
|
|
|
| def list_processed(proc_dir): |
| cases = [os.path.splitext(os.path.basename(p))[0] |
| for p in glob.glob(os.path.join(proc_dir, "*.npz"))] |
| return sorted(cases, key=numeric_id) |
|
|
|
|
| def _in(num, rng): |
| return bool(rng) and rng[0] <= num <= rng[1] |
|
|
|
|
| def make_split(proc_dir, cfg): |
| cases = list_processed(proc_dir) |
| sp = cfg["split"] |
| exclude_cases = set(sp.get("exclude_cases", []) or []) |
| train, test = [], [] |
| for c in cases: |
| n = numeric_id(c) |
| if n in exclude_cases: |
| continue |
| if _in(n, sp.get("exclude_range")): |
| continue |
| if _in(n, sp.get("test_range")): |
| test.append(c) |
| elif _in(n, sp.get("train_range")): |
| train.append(c) |
| return train, test |
|
|
|
|
| def kfold(train_cases, n_folds=5, seed=42): |
| rng = np.random.default_rng(seed) |
| idx = np.arange(len(train_cases)) |
| rng.shuffle(idx) |
| folds = np.array_split(idx, n_folds) |
| out = [] |
| for i in range(n_folds): |
| val_idx = folds[i] |
| tr_idx = np.concatenate([folds[j] for j in range(n_folds) if j != i]) \ |
| if n_folds > 1 else val_idx |
| out.append(([train_cases[j] for j in tr_idx], |
| [train_cases[j] for j in val_idx])) |
| return out |
|
|
