Adịghị nhatanha na klaasị
Class imbalance is when one outcome vastly outnumbers another — like 99.9% legitimate transactions versus 0.1% fraud — which tricks models into ignoring the rare but important class.
Nchịkọta
Resampling rebalances the training data so the model actually learns to spot the minority.
Ime miri emi
Mgbe klaasị na-agbagọ, ihe nlereanya nwere ike iru 99.9% izi ezi site n'ịkọ ọnụ ọgụgụ ka ukwuu na mgbe ọ bụla ijide otu wayo, nke na-abaghị uru. Resampling na-edozi nkesa ọzụzụ n'ụzọ abụọ sara mbara. Ntugharị nke oyiri ma ọ bụ na-ejikọta ihe atụ pere mpe - SMOTE kpochapụwo (Synthetic Minority Over-Sampling Technique) na-emepụta isi ihe ọhụrụ site na itinye ọnụ n'etiti obere ihe nlele na ndị agbata obi ya kacha nso kama iṅomi ha. Ịghọta kama ịtụfu ọtụtụ ihe atụ (na-enweghị usoro, ma ọ bụ nke ọma site na ụzọ dị ka njikọ Tomek ma ọ bụ NearMiss) na-eme ka ihe pụta ìhè, na-efu nke ịtụfu data. Nhọrọ ndị ọzọ na-ezere imetụ data ahụ gụnyere ịdị arọ klaasị (ịnata mmejọ pere mpe karịa na arụ ọrụ ọnwụ) na imezi oke mkpebi mgbe ọzụzụ gasịrị.
Nghọta nka nka
Iwu dị oke egwu: chegharịa naanị usoro ọzụzụ, ọnweghị nkwado ma ọ bụ setịpụ ule, ma na-emegharịgharị mgbe niile n'ime mpịachi nkwenye. Ịmebiga ihe ókè tupu ikewa ihie ụzọ dị nso-abụọ n'ime ihe nlele ule ma na-ebuli akara. N'ihi na izi ezi enweghị isi ebe a, nyocha kwesịrị ịdabere na nkenke, cheta, F1, Precision-Recall AUC, ma ọ bụ Matthews Correlation Coefficient - metrik na-akwụwa aka ọtọ mgbe klaasị dị mma adịghị ụkọ.
Mmetụta atụmatụ
Ọnụ ego na mmefu ego
Mkpebi ihe owuwu ụlọ na-akwalite arụmọrụ yana ọnụ ahịa ọrụ ruo ọtụtụ afọ.
Mkpebi doro anya
Nkà mmụta nka na-enyere ndị otu egwuregwu aka ịhọrọ nchịkọta ziri ezi, ọ bụghị naanị nke kachasị ọhụrụ.
Quality akara
Nhọrọ injinia ka mma na-ebelata ihe omume ntụkwasị obi na mmepụta.
Ọdịnihu nke klaasị na nhazigharị
Nhazigharị na-esiwanye ike n'ime pipeline ML, yana ụlọ akwụkwọ ọbá akwụkwọ dị ka mmụta na-ezighi ezi na-ejikọta ozugbo na nkwado ndabere. Nchọpụta na-aga n'ihu n'ịmụ ihe na-efu ọnụ yana ọrụ mfu ahaziri ahazi - dị ka mfu isi, nke na-agbadata ọtụtụ ihe atụ dị mfe - nke na-egosipụtakarị nrụpụta crude na netwọk miri emi. Maka data nchịkọta akụkọ na onyonyo, ụdị mmepụta nke na-emepụta ihe nlele pere mpe na-apụta dị ka onye ga-anọchi anya ọkaibe na njikọta ụdị SMOTE.
Mmejuputa n'ezie n'ụwa
Ọzụzụ ihe nchọpụta wayo kaadị kredit ebe ezigbo aghụghọ dị n'okpuru 1% nke azụmahịa, na-eji SMOTE kwalite okwu wayo na-adịghị ahụkebe.
Ịmepụta usoro ahụike maka ọrịa na-adịghị ahụkebe dị naanị na pasent ole na ole nke ndị ọrịa, na-etinye ọnụ ọgụgụ klaasị ka a na-ata ahụhụ nke ukwuu.
Ịchọta ihe ndị nwere ntụpọ n'ahịrị nrụpụta ebe ihe fọrọ nke nta ka ọ bụrụ ngwaahịa niile na-agafe nyocha, na-elele ihe 'dị mma' iji dozie ọzụzụ.
Na-edepụta ntinye netwọọk na-adịghị ahụkebe na ndekọ cybersecurity nke okporo ụzọ nkịtị na-achịkwa, jiri Precision-Recall AUC enyocha ya kama izi ezi.
Ihe ize ndụ & okporo ụzọ nche
Ịkwalite otu akara ngosi nwere ike zoo adịghị ike sistemụ sara mbara.
A na-eledakarị ihe akụrụngwa na ụgwọ ọrụ anya.
Ọdịiche nchekwa na nleba anya nwere ike itolite ka sistemu na-adịwanye mgbagwoju anya.
Map mmejuputa
Kọwaa latency, ịdịmma na ebumnuche ọnụ ahịa tupu mmejuputa ya.
Benchmark n'okpuru ibu dị adị na ọnọdụ data.
Nleba anya akụrụngwa maka mperi, ịkpafu na mmetụta onye ọrụ.
Kwadebe ụzọ nzaghachi azụghachi azụ na ihe omume tupu ịchachaa.
Nọgide na-eme nchọpụta
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Class Imbalance and Resampling quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Ntuziaka na-esote
Netwọk Siamese na mfu Triplet
Ajụjụ a na-ajụkarị
What is Class Imbalance and Resampling?
Ahaghị nhata klaasị bụ mgbe otu nsonaazụ karịrị nke ọzọ - dị ka azụmahịa 99.9% ziri ezi na aghụghọ 0.1% - nke na-aghọ aghụghọ n'ụdị ileghara klaasị dị ụkọ mana dị mkpa. Nhazigharị na-emegharị data ọzụzụ ka ihe nlereanya ahụ mụta n'ezie ịhụ ndị pere mpe.
Kedu ihe kpatara izi ezi doro anya ji bụrụ metrik na-adịghị mma maka nsogbu nhazi ọkwa enweghị oke?
Ọ bụrụ na 99.9% nke ikpe na-adịghị mma, ihe nlereanya na-ebu amụma mgbe niile na-enweta izi ezi 99.9% mgbe ọ na-enweta ihe efu efu, yabụ izi ezi na-ezochi ọdịda zuru oke na klaasị.
Kedu ihe SMOTE na-eme?
SMOTE (Synthetic Minority Over-Sampling Technique) na-emepụta ihe atụ ọhụrụ pere mpe site n'itinye ọnụ n'etiti ebe pere mpe na ndị agbata obi ya kacha nso, kama ịmegharị ha naanị.
Kedu ụzọ na-edozi enweghị ahaghị nhata na-agbanweghị ọnụọgụ ihe atụ ọzụzụ?
Ịdị arọ klaasị na-ahapụ data ahụ emetụghị aka kama na-eme ka ọrụ mfu na-ata ahụhụ na klaasị pere mpe karịa.
Kedu ihe bụ isi ihe egwu dị n'itinye nleba anya tupu ị kewaa data gị n'ime ụgbọ oloko na ule ule?
Ịmegharị tupu nkewa ahụ na-eme ka isi ihe dị nso na-apụta na ụgbọ oloko na n'ule ule, na-ewepụta ozi ma na-emepụta nchekwube gabigara ókè, arụmọrụ na-enweghị isi.
Gịnị bụ isi ọdịda nke random undersampling?
Ịghọta klaasị na-edozi klaasị site na ịtụfu ọtụtụ ihe atụ, nke nwere ike tụfuo data na-enye ihe ọmụma ma mebie ikike onye nlereanya nwere ịmụta klaasị.