HejuruUbuyobozi bukurikira
Kwibuka, Ubunyamateka n'Igerageza ry'Ukunesha mu Bugenzuzi bw'Inyandiko
Tekiniki
Ubuyobozi bwa tekiniki
Ubusanzwe bw'ukuri bugaragaza ukuri kwa model uko kwibuka kwiyongera mu ntera z'ibyemezo.
Umva iki gitaboAI-byavuzwe · ~3 min
Imbonerahamwe z'ubushobozi bwo kwemeza n'ubushobozi bwo kwemeza zifite akamaro iyo ibimenyetso by'ukuri biba bike, ariko ubunini bw'ubushobozi bwo kwemeza n'ubushobozi bwo kwemeza bw'ubushobozi bwo kwemeza ni uburyo butandukanye bukwiye kwitwa neza.
Precision asks how many selected cases are positive. Recall asks how many of all positive cases have been selected. As a model's score threshold is lowered, more cases enter the selected set. Recall cannot decrease as the threshold is lowered, but precision can rise or fall depending on the newly included cases. A precision-recall curve traces this tradeoff. It is especially informative for a rare positive class because it directly exposes how many selected items are false alarms. A receiver operating characteristic curve answers a different question using the false-positive rate, whose denominator includes all actual negatives. Neither view removes the need to consider real decision costs. Average precision summarizes the precision-recall relationship by weighting precision values by the corresponding increases in recall. In a simple ranking without tied scores, it equals the average precision at the ranks occupied by positive cases. Suppose the only two positives appear first and third. The precisions at those positions are one and two-thirds, so average precision is about 0.833. The label PR-AUC can be ambiguous. Some reports mean trapezoidal integration of precision against recall; others use it loosely for average precision. Scikit-learn distinguishes average_precision_score from auc, which applies the trapezoidal rule. These values can differ because the interpolation assumptions differ. State the actual calculation. Precision depends on the prevalence of positives in the evaluation population. A score from a balanced sample may not describe a deployment setting where positives are rare. Report the positive fraction and evaluation design. Finally, a summary across thresholds does not select an operating threshold. Inspect the precision, recall and workload at the point where the model will actually be used.
Urutonde rw'isesengura rufite ibintu byinshi bisanzwe n'ibintu bike bifite akamaro. Ikigereranyo cya precision-recall kigaragaza ubusabane hagati yo kubona ibintu bifite akamaro byinshi no gusaba abasesenguzi gusuzuma ibitari ngombwa.
Mu ishusho y'irangamuntu itagira amanota angana, ibiri byiza biboneka bwa mbere n'icya gatatu. Precision kuri izo myanya ni 1 na kabiri, bityo ubusanzwe bw'ukuri buri hafi 0.833.
Itsinda rikoreshwa rya scikit-learn's average_precision_score ku bipimo by'ibimenyetso n'ibipimo by'ibisubizo. Rerekana urwego rw'ibimenyetso by'ukuri kandi ritakaza gusimbuza ibisubizo n'ibisubizo by'ukuri mbere yo kugenzura.
Ibyiciro bibiri by'ibipimo by'ibisubizo birimo ibipimo bitandukanye by'ibimenyetso by'ukuri. Umukozi w'ibipimo asubiza ibyo bipimo mbere yo kugereranya ibisubizo by'ubushobozi bwo kwemeza, kuko prevalence ihindura uko ubunini bw'ubushobozi bwo kwemeza bugomba gusobanurwa.
Dashboard z'isesengura zishobora gutuma ibipimo by'amanota birushaho kugaragara neza binyuze mu kwerekana ubwinshi bw'ibintu byiza, ibarurishamibare nyakuri y'ubuso n'ahantu nyakuri ho gukorera hafi y'amanota y'ingenzi. Itsinda rishinzwe isuzuma rishobora kwita cyane ku mubare w'ibibazo bifite akamaro bigaragara mu bushobozi bwa buri munsi bwagenwe, mu gihe itsinda ry'indi rishobora gukenera igipimo gito cya recall. Gukomeza kubika ibyahanuwe n'amanota bituma ibibazo nk'ibi bishobora gusubirwamo nta guhindura model. Mu gihe abaturage bahinduka, amatsinda agomba kongera gusuzuma ubusumbane ku makuru ahagarariye aho kwemeza ko umusaruro w'ubwiza bw'ibanze (average-precision) w'igihe cyashize wemeza umubare w'akazi cyangwa akamaro mu gihe kizaza.
Ibaruwa y'amakuru y'ubuntu
Inkuru eshatu zizewe za AI buri gitondo cy'iminsi y'akazi, zanditse mu rurimi rworoshye. Ubuntu iteka ryose, nta kwamamaza.
Imeri imwe buri munsi y'akazi. Kwiyimura mu buryo bumwe. Ntitwigera tugurisha cyangwa ngo dusangire aderesi yawe.
Isuzume
Ibisubizo ako kanya kuri buri kibazo, n'icyemezo gishobora gusangirwa gifite ID yagenzurwa umaze gutsinda isomo.
Fasha uburezi bwa AI bw'ubuntu. AI Understanding ni umuryango utegamiye kuri leta wa 501(c)(3) — nta kwamamaza, nta kwishyura, iteka ryose. Tanga impano
Precision y'ubusanzwe isobanura ubunyangamugayo bw'umodeli uko recall yiyongera ku ntera z'ibyemezo. Imirongo ya precision-recall ifite akamaro iyo ibibazo byiza ari bike, ariko precision y'ubusanzwe n'ubuso bwa trapezoidal munsi y'iyo mirongo ni imibare itandukanye ikwiye kwitwa neza.
Precision ipima isuku y'itsinda ryatoranijwe; recall ipima igice cy'ibyo byiza byose byabonetse.
Precision ni imwe ku ya mbere y'ibya neza kandi igice bibiri ku ya kabiri. Impuzandengo yabyo ni igice cya gatanu ku cyenda, hafi 0.833.
Precision y'ubusanzwe ibara precision hakurikijwe increments za recall, mu gihe integration ya trapezoidal ihuza ibipimo by'imirongo ikoresheje uburyo butandukanye bw'ubuso.
Amanota atuma habaho isuzuma ku ntera z'ibyemezo. Ibyemezo bikomeye bibika gusa ahantu kamwe ko gukora kandi bigasiba amakuru menshi y'urutonde.
Precision ishingira ku bwinshi bw'icyiciro, bityo abaturage n'uburyo bwo gufata ingero ni ibikenewe mu context.
Komeza wige
Abandi bayobozi batoranijwe kuriyi ngingo
HejuruUbuyobozi bukurikira
Kwibuka, Ubunyamateka n'Igerageza ry'Ukunesha mu Bugenzuzi bw'Inyandiko
Tekiniki