SelanjutnyaPanduan berikutnya
Validasi Silang Bersarang
Teknis
PANDUAN Dasar
Cross-validation estimates how a model or modeling procedure performs on held-out data by repeating train/validation splits.
Dengarkan panduan iniDiceritakan oleh AI · ~1 min
It supports model assessment, but the estimate depends on the data, split design and preprocessing.
In k-fold cross-validation, the data are divided into folds; each fold is held out once while the others train the model. Repeated partitions can make an estimate less dependent on one particular split, but do not guarantee greater reliability: dependence between folds, sample size and distribution shift still matter. Preprocessing belongs inside each training fold. Nested cross-validation separates hyperparameter selection from outer evaluation.
K-fold cross-validation rotates which fold is held out for evaluation.
Stratification can preserve approximate class proportions in folds for classification.
Preprocessing should be fitted within each training fold to avoid leakage.
Nested cross-validation separates inner model selection from outer evaluation.
Choose validation procedures based on the data structure and deployment question. AutoML or deep-learning workflows do not universally require one holdout strategy; the method should be justified and reported.
Buletin gratis
Tiga cerita AI terverifikasi setiap pagi hari kerja, ditulis dalam bahasa Inggris sederhana. Gratis selamanya, tanpa iklan.
Satu email setiap hari kerja. Berhenti berlangganan dalam satu klik. Kami tidak pernah menjual atau membagikan alamat Anda.
Uji diri Anda sendiri
Umpan balik instan untuk setiap jawaban, dan sertifikat yang dapat dibagikan dengan ID yang dapat diverifikasi setelah Anda lulus kursus.
Mendukung pendidikan AI gratis. AI Understanding adalah organisasi nirlaba 501(c)(3) — tanpa iklan, tanpa paywall, selamanya. Berikan sumbangan
Cross-validation estimates how a model or modeling procedure performs on held-out data by repeating train/validation splits. It supports model assessment, but the estimate depends on the data, split design and preprocessing.
Setiap lipatan berfungsi sebagai set pengujian tepat satu kali dalam putaran k, sehingga setiap sampel diuji satu kali dan dilatih pada sisanya.
Averaging results across folds can reduce dependence on one particular train/test split, but it does not guarantee a reliable estimate; sample size, data dependence and split design still matter.
Mengacak data berdasarkan waktu akan membocorkan masa depan ke dalam pelatihan, sehingga CV deret waktu menggunakan pemisahan rantai maju yang hanya melatih observasi sebelumnya.
Nested cross-validation uses inner folds for model selection and outer folds to estimate the performance of that selection process, reducing selection bias.
The guide identifies five- and ten-fold cross-validation as common choices, though the suitable design depends on the data and task.
Teruslah belajar
Panduan lainnya dipilih untuk topik ini
SelanjutnyaPanduan berikutnya
Validasi Silang Bersarang
Teknis