Ka wahahee anaana

Ka Bias of Analysis

Ua ʻōlelo ʻo Mark Twain i kekahi mea e like me, "ʻEkolu ʻano o ka wahaheʻe: ka wahaheʻe, ka wahaheʻe a me ka hewa. analytics. " 

Manaʻo mākou e hāʻawi ana nā analytics iā mākou i nā ʻike pono a hiki ke hana. ʻO ka mea a mākou i ʻike ʻole ai ʻo ia ke ʻano o kā mākou mau manaʻo ponoʻī a me nā mea ʻē aʻe e hoʻohuli ai i nā pane i hāʻawi ʻia iā mākou e nā polokalamu a me nā ʻōnaehana ʻoi loa. I kekahi manawa, hiki ke hoʻopunipuni ʻia mākou me ka hoʻopunipuni, akā, ʻoi aku ka maʻamau, he mau manaʻo maʻalahi a ʻike ʻole paha ia e komo i loko o kā mākou ʻikepili. ʻO ka hoʻoikaika ʻana ma hope o ka ʻikepili biased he nui. I kekahi manawa, ua hoʻololi ʻia nā hopena ʻokoʻa a mākou e manaʻo ai mai ka ʻepekema e 1) nā koho maʻalahi i ke ʻano o ka hōʻike ʻana i ka ʻikepili, 2) ʻike like ʻole a i ʻole ka ʻike makamaka, 3) pehea e aʻo ʻia ai nā ʻōnaehana AI, 4) ka naʻaupō, ka hemahema o nā mea noiʻi a i ʻole nā ​​​​mea ʻē aʻe e hoʻāʻo nei. e haʻi i ka moʻolelo, 5) ka hoʻopaʻa ʻana iā ia iho.    

Kūlike ka Hōʻike

ʻOi aku ka maʻalahi o kekahi o nā wahaheʻe ma mua o nā mea ʻē aʻe. Ke ʻike ʻoe i ka mea e ʻimi ai e ʻike maʻalahi paha ʻoe kiʻi hoʻopunipuni a me nā pakuhi. 

Aia ma ka liʻiliʻi ʻelima ala e hōʻike hewa ai i ka ʻikepili: 1) Hōʻike i kahi pūʻulu ʻikepili palena ʻole, 2). Hōʻike i nā pilina pili ʻole, 3) Hōʻike pololei ʻole i ka ʻikepili, 4) Hōʻike i ka ʻikepili ma ke ʻano maʻamau, a i ʻole 5). Hōʻike i ka ʻikepili i maʻalahi.

Hōʻike i kahi pūʻulu ʻikepili palena ʻole

ʻO ka hoʻopaʻa ʻana i ka ʻikepili, a i ʻole ke koho lima ʻana i kahi ʻāpana manuahi ʻole o ka ʻikepili hiki ke haʻi pinepine i kahi moʻolelo i kūlike ʻole me ke kiʻi nui. ʻO ka hōʻiliʻili maikaʻi ʻole, a i ʻole ka ʻohi cherry, ʻo ia ka manawa e hoʻohana ai ka mea loiloi i kahi laʻana mea ʻole e hōʻike i kahi hui nui. 

Ma Malaki 2020, Ke Keʻena Ola Aupuni o Georgia paʻi ʻia kēia pakuhi ma ke ʻano o kāna hōʻike kūlana i kēlā me kēia lā. ʻOi aku ka nui o nā nīnau ma mua o ka pane ʻana.  

ʻO kekahi o nā mea i nalowale ʻo ia ka pōʻaiapili. No ka laʻana, he mea kōkua ke ʻike i ka pākēneka o ka heluna kanaka no kēlā me kēia pūʻulu makahiki. ʻO kekahi pilikia me ka pakuhi paʻi maʻalahi ke nānā aku ʻo ia nā hui makahiki ʻole. He 0 makahiki ka 17-18, he 18 ka 59-42, he 60+ ka wehe ʻana, akā ma kahi o 40 mau makahiki. ʻO ka hopena, hāʻawi ʻia i kēia pakuhi wale nō, ʻo ka hapa nui o nā hihia i loko o ka pūʻulu makahiki 18-59 makahiki. ʻO ka hui o 60+ mau makahiki he mea liʻiliʻi loa ia e nā hihia COVID. Akā ʻaʻole kēia ka moʻolelo holoʻokoʻa.

No ka hoʻohālikelike, kēia ʻikepili ʻokoʻa i hoʻonohonoho ʻia ma ka Paena pūnaewele CDC palapala i nā hihia COVID e ka pūʻulu makahiki me ka ʻikepili hou e pili ana i ka pākēneka o ka heluna kanaka US ma kēlā me kēia pae makahiki.  

ʻOi aku ka maikaʻi o kēia. Loaʻa iā mākou ka pōʻaiapili. Hiki iā mākou ke ʻike i nā hui makahiki 18-29, 30-39, 40-49 ka nui o nā hihia ma mua o ka pākēneka o ka pūʻulu makahiki i ka heluna kanaka. Aia nō kekahi mau pūʻulu makahiki like ʻole. No ke aha he hui makahiki kaawale ka 16-17? ʻAʻole naʻe kēia ka moʻolelo holoʻokoʻa, akā ua kākau nā poʻe loea i nā kolamu, hana i nā wānana a me nā kauoha ma lalo o kēia. ʻOiaʻiʻo, me COVID, nui nā ʻano like ʻole i ka makahiki e pili ana i ka helu ʻia ʻana ma ke ʻano he hihia kūpono: ke kūlana kano, ka loaʻa ʻana o nā hoʻokolohua, ka nui o nā manawa i hoʻāʻo ʻia, comorbidities, a me nā mea ʻē aʻe he nui. ʻO ka helu o nā hihia, ʻo ia iho, hāʻawi i kahi kiʻi piha ʻole. ʻO ka hapa nui o nā poʻe loea e nānā i ka helu o nā make, a i ʻole nā ​​​​pakeneka o nā make ma 100,000 heluna, a i ʻole nā ​​hihia-fatalities e nānā i ka hopena o COVID i kēlā me kēia makahiki.

Hōʻike i nā pilina pili ʻole

ʻIke loa, aia kahi pili ikaika ma waena o ka hoʻolilo ʻana o ʻAmelika i ka ʻepekema, ka lewa, a me ka ʻenehana a me ka nui o nā mea pepehi kanaka ma ke kau ʻana, ʻoki ʻia a me ka hoʻopaʻa ʻana. ʻO ka Correlation he 99.79%, kokoke i kahi hoʻokūkū kūpono.  

ʻO wai naʻe, e hana i ka hihia e pili ana kēia mau mea, a i ʻole kekahi kumu i kekahi? Aia kekahi mau hiʻohiʻona liʻiliʻi liʻiliʻi, akā ʻaʻole liʻiliʻi i ka spurious. Aia kekahi pilina ikaika like ma waena o nā Letters in Winning Word of Scripps National Spelling Bee a me ka helu o nā kānaka i pepehi ʻia e Venomous Spiders. ʻAkahi? Hoʻoholo ʻoe.

ʻO kekahi ala ʻē aʻe e kiʻi i kēia ʻikepili i emi iki paha ka hoʻopunipuni ʻana ʻo ia ka hoʻokomo ʻana i ka zero ma nā koʻi Y ʻelua.

Hōʻike pololei ʻole ka ʻikepili

mai ka Pehea e hōʻike hewa ai i ka ʻikepili, ua hōʻike ka US State of Georgia i nā ʻāpana 5 kiʻekiʻe me ka helu nui loa o nā hihia COVID-19 i hōʻoia ʻia.

Nānā pono, ʻeā? Aia ke ʻano i lalo o nā hihia COVID-19 i hoʻopaʻa ʻia. Hiki iā ʻoe ke heluhelu i ka X-axis? Hōʻike ka X-axis i ka manawa. ʻO ka maʻamau, e hoʻonui ʻia nā lā mai ka hema a i ka ʻākau. Ma ʻaneʻi, ʻike mākou i kahi huakaʻi manawa liʻiliʻi ma ka X-axis: 

4/28/2020

4/27/2020

4/29/2020

5/1/2020

4/30/2020

5/4/2020

5/6/2020

5/5/2020

5/2/22020 ...

E kali? He aha? ʻAʻole hoʻokaʻawale ʻia ka axis X i ka manawa. No laila, e like me ka maikaʻi o ke ʻano, ʻaʻole hiki iā mākou ke huki i nā hopena. Inā kauoha ʻia nā lā, ʻoi aku ka nui o ke ʻano o ka sawtooth ma mua o ke ʻano o nā ʻano.

ʻO ka hoʻoponopono maʻalahi maʻaneʻi ʻo ka hoʻokaʻawale ʻana i nā lā e like me ka hana ʻana o ka kalena.

Hōʻike i ka ʻikepili me ka ʻole

Paʻa mākou a pau. Ua aʻo ko mākou lolo iā mākou e hoʻoholo wikiwiki e pili ana i nā manaʻo i kūlike i ko mākou honua. No ka laʻana, ʻo kēlā me kēia pakuhi aʻu i ʻike ai e hōʻike ana i ka hui ʻana o nā koʻi x a me y i ʻole, a i ʻole nā ​​kumu waiwai haʻahaʻa. Ke nānā pōkole nei i kēia pakuhi, he aha nā hopena āu e huki ai e pili ana i ka hopena o Florida “E kū i kāu kānāwai kumu.?? Hilahila au ke ae aku, aka, ua hoopunipuni keia pakuhi ia'u i ka mua. Hiki ke huki ʻia kou maka i ke kikokikona a me ka pua i waenakonu o ke kiʻi. Aia i lalo ma kēia pakuhi. ʻAʻole paha he wahaheʻe - aia ka ʻikepili ma laila. Akā, pono wau e noʻonoʻo he mea hoʻopunipuni ia. Inā ʻaʻole ʻoe i ʻike iā ia, aia ka zero ma ke axis y ma luna. No laila, i ka emi ʻana o ka ʻikepili, ʻoi aku ka nui o nā make. Hōʻike kēia pakuhi i ka nui o nā pepehi kanaka e hoʻohana ana i nā pū ua hoʻonuiʻia ma hope o 2005, hōʻike ʻia e ke ʻano hele iho.

Hōʻike i ka ʻikepili maʻalahi

Hiki ke ʻike ʻia kekahi hiʻohiʻona o ka maʻalahi o ka ʻikepili i ka wā e hoʻohana pono ai nā mea loiloi iā Simpson's Paradox. He hanana kēia i ka wā i ʻike ʻia ai ka ʻikepili i hōʻuluʻulu ʻia e hōʻike i kahi hopena ʻokoʻa ma mua o ka wā i hoʻokaʻawale ʻia i nā ʻāpana. He maʻalahi kēia pahele i ka nānā ʻana i nā pākēneka hōʻuluʻulu kiʻekiʻe. ʻO kekahi o nā kiʻi maopopo loa o Simpson's Paradox ma ka hana e pili ana i pāʻani awelika.  

Ma ʻaneʻi mākou e ʻike ai he ʻoi aku ka ʻoi aku o ka awelika batting ma mua o David Justice no nā kau 1995 a me 1996. Hele mai ka paradox i ka wā e ʻike ai mākou ua lanakila ʻo Justice iā Jeter i ka pāʻani ʻana i kēlā mau makahiki ʻelua. Inā ʻoe e nānā pono, kūpono ke ʻike ʻoe ua loaʻa iā Jeter he 4x ʻoi aku ka nui o nā ʻōpeʻapeʻa (ka denominator) i ka makahiki 1996 ma kahi .007 haʻahaʻa haʻahaʻa i ka makahiki 1996. ʻOiai, ua loaʻa iā Justice ma kahi o 10x ka nui o nā ʻōpeʻa ma . 003 kiʻekiʻe awelika ma 1995.

Hōʻike pololei ʻia ka hōʻike, akā ʻo Simpson's Paradox, me ka ʻike ʻole, a me ka ʻike ʻole paha, ua alakaʻi i nā hopena hewa. I kēia mau lā, aia kekahi mau hiʻohiʻona o Simpson's Paradox i ka nūhou a me ka media media e pili ana i nā kano a me ka make COVID. ʻekahi pakuhi hōʻike i kahi pakuhi laina e hoʻohālikelike ana i nā helu make ma waena o ka poʻe i hoʻopaʻa ʻia a me ka poʻe i hoʻopaʻa ʻole ʻia no ka poʻe 10-59 makahiki. Hōʻike ka pakuhi he haʻahaʻa ka nui o ka make o ka poʻe i hoʻopaʻa ʻole ʻia. He aha ka hana ma ʻaneʻi?  

Ua like ka pilikia me ka mea a mākou e ʻike nei me nā awelika batting. ʻO ka denominator i kēia hihia ka helu o nā kānaka i kēlā me kēia pūʻulu makahiki. Hoʻohui ka pakuhi i nā hui i loaʻa nā hopena like ʻole. Inā mākou e nānā i ka poʻe ʻelemakule, 50-59, ʻoi aku ka maikaʻi. Pēlā nō, inā mākou e nānā i ka 10-49, ʻike mākou ʻoi aku ka maikaʻi o ka huakaʻi. ʻO ka paradoxically, i ka nānā ʻana i ka hui i hui ʻia, ʻike ʻole ʻia ka hopena maikaʻi ʻole. Ma kēia ala, hiki iā ʻoe ke hana i kahi hihia no nā hoʻopaʻapaʻa kū'ē me ka hoʻohana ʻana i ka ʻikepili.

Kūlike ka ʻikepili

ʻAʻole hiki ke hilinaʻi ʻia ka ʻikepili i nā manawa a pau. ʻOiai ma ke kaiāulu ʻepekema, ʻoi aku ka hapakolu o nā mea noiʻi i noiʻi ʻia "nā hana noiʻi nīnau."  kekahi noiʻi hoʻopunipuni makaʻu 'Ōlelo mai, "He nui aku paha ka hoʻopunipuni i loko o ka ʻikepili - nā papa, nā laina laina, ka ʻikepili hoʻonohonoho [- ma mua o kā mākou e ʻike maoli nei]. Hiki i nā mea a pau e noho ana ma kā lākou papaʻaina kīhini ke kau i kekahi mau helu i loko o ka pāpalapala a hana i ka pakuhi laina e ʻike maopopo ʻia.

ʻO kēia mua Eia me he mea lā ua hana kekahi. ʻAʻole wau e ʻōlelo nei he hoʻopunipuni kēia, akā ma ke ʻano he noiʻi, ʻaʻole ia e hoʻopuka i kekahi ʻikepili e kōkua i ka hoʻoholo ʻike. Me he mea lā ua nīnau ka noiʻi i ka poʻe pane e pili ana i ko lākou manaʻo no ke kofe hoʻokele kinoea, a i ʻole kekahi hanana e pili ana i kēia manawa. 

1. ʻoi loa 
2. Nui
3. Maikaʻi loa 

Ua ʻoki au i ka pou Twitter no ka wehe ʻana i nā kuhikuhi i ka ʻaoʻao hewa, akā ʻo kēia ka pakuhi piha maoli o nā hopena hope o ka noiʻi. ʻAʻole maʻamau nā noiʻi e like me kēia. ʻIke loa, ʻo kēlā me kēia pakuhi i hana ʻia mai ka ʻikepili i loaʻa mai nā pane e hōʻike i ke kofe i nīnau ʻia ʻaʻole e poina.  

ʻO ka pilikia, inā ua hāʻawi ʻia kēia noiʻi iā ʻoe a ʻaʻole i loaʻa kahi pane i kūpono i kou noʻonoʻo, e hoʻōki ʻoe i ka loiloi. He laʻana koʻikoʻi paha kēia o ka hana ʻana i ka ʻikepili hilinaʻi ʻole. ʻO ka hoʻolālā noiʻi maikaʻi ʻole, hiki ke alakaʻi i ka liʻiliʻi o nā pane a hoʻokahi wale nō manaʻo o ka poʻe pane, ʻo ia wale nō ka manaʻo. Kūlike ka ʻikepili.

ʻO kēia hiʻohiʻona lua o ka bias data mai nā faila o "ʻO nā kiʻi hoʻopunipuni COVID 19 maikaʻi loa. " 

Eia hou, maʻalahi kēia a ʻaʻole maopopo loa. Hōʻike ka pakuhi pahu i kahi maʻalahi - kokoke i ka maʻalahi - ka emi ʻana o ka pākēneka o nā hihia COVID-19 maikaʻi i ka manawa no kahi kalana ma Florida. Hiki iā ʻoe ke huki i ka manaʻo e emi ana nā hihia. Maikaʻi kēlā, hōʻike pololei ka ʻike i ka ʻikepili. Aia ka pilikia ma ka ʻikepili. No laila, ʻoi aku ka maikaʻi o ka manaʻo no ka mea ʻaʻole hiki iā ʻoe ke ʻike. Ua kālua ʻia i ka ʻikepili. ʻO nā nīnau āu e nīnau ai, ʻo wai ka mea e hoʻāʻo ʻia? ʻO ia hoʻi, he aha ka denominator, a i ʻole ka heluna kanaka a mākou e nānā nei i ka pākēneka. ʻO ka manaʻo, ʻo ia ka heluna kanaka holoʻokoʻa, a i ʻole ka liʻiliʻi, he hōʻailona hōʻike.

Eia nō naʻe, i kēia manawa, ma kēia kalana, hāʻawi ʻia nā hoʻokolohua i kahi helu palena ʻole o ka poʻe. Ua loaʻa iā lākou nā hōʻailona like me COVID, a i ʻole ua huakaʻi hou i kahi ʻāina ma ka papa inoa o nā wahi wela. Hoʻohui hou i nā hopena ʻo ia ka helu ʻana i kēlā me kēia hōʻike maikaʻi a helu ʻia kēlā me kēia hōʻike maikaʻi ʻole. ʻO ka maʻamau, i ka wā i hoʻāʻo ai ke kanaka i ka maikaʻi, e hoʻāʻo hou lākou i ka wā i holo ai ka maʻi i kona ala a e hoʻāʻo maikaʻi ʻole. No laila, ma ke ʻano, no kēlā me kēia hihia maikaʻi, aia kahi hihia hoʻāʻo maikaʻi ʻole e kāpae iā ia. ʻO ka hapa nui o nā hoʻāʻo he maikaʻi ʻole a helu ʻia nā hoʻokolohua maikaʻi ʻole o kēlā me kēia kanaka. Hiki iā ʻoe ke ʻike i ke ʻano o ka ʻikepili a ʻaʻole pono no ka hoʻoholo ʻana. 

ʻO ka hoʻokomo ʻana a me ka hoʻomaʻamaʻa ʻana o AI ʻo Biased

Aia ma lalo o ʻelua mau ala e hiki ai iā AI ke alakaʻi i nā hopena ʻino: hoʻomaka me ka ʻikepili pili, a i ʻole ka hoʻohana ʻana i nā algorithm biased e hana i ka ʻikepili kūpono.  

Kuhikuhi Kūlike

He nui ka poʻe o mākou ma lalo o ka manaʻo e hiki ke hilinaʻi ʻia ʻo AI e hoʻopau i nā helu, e hoʻopili i kāna mau algorithms, a kuha i kahi loiloi hilinaʻi o ka ʻikepili. Hiki ke akamai wale ka Artificial Intelligence e like me ka mea i aʻo ʻia. Inā kūpono ʻole ka ʻikepili i aʻo ʻia ai, ʻaʻole hiki ke hilinaʻi ʻia nā hopena a i ʻole nā ​​hopena. E like me ka hihia ma luna o ka noiʻi bias, aia kekahi mau ala e hiki ai i ka ʻikepili makemakeʻia i ka mīkini aʻo:.  

• Laʻana hoʻohālikelike - ʻaʻole hōʻike ka ʻikepili aʻo i ka heluna kanaka holoʻokoʻa.
• Hoʻokaʻawale ʻia - i kekahi manawa ʻike maoli ʻia nā mea i waho, a i ʻole, kahi mākou e huki ai i ka laina i ka mea e hoʻokomo ai (zip code, nā lā, etc.).
• ʻO ke ana ʻana - pono ke ana ʻana mai ke kikowaena a me ka lalo o ka meniscus, no ka laʻana, ke ana ʻana i nā wai i loko o nā ipu volumetric a i ʻole nā ​​paipu hoʻāʻo (koe wale ka mercury.)
• E hoʻomanaʻo i ka manaʻo ʻino – ke hilinaʻi ka noiʻi i ka hoʻomanaʻo o nā mea komo.
• Kūleʻa nā mea nānā - ʻo nā ʻepekema, e like me nā kānaka āpau, ʻoi aku ka makemake e ʻike i nā mea a lākou e manaʻo ai e ʻike.
• Hoʻokaʻawale i ke kāne a me ka hoʻokalakupua - hiki ke hōʻike ʻia ka wahine a i ʻole ka lāhui.  
• ʻAha hui - hoʻoikaika ka ʻikepili i nā stereotypes

No ka hoʻihoʻi ʻana o AI i nā hopena hilinaʻi, pono kāna ʻikepili aʻo e hōʻike i ka honua maoli. E like me kā mākou i kūkākūkā ai ma kahi moʻomanaʻo moʻomanaʻo ma mua, he mea koʻikoʻi ka hoʻomākaukau ʻana i ka ʻikepili a like me nā papahana ʻikepili ʻē aʻe. Hiki i nā ʻikepili hilinaʻi ʻole ke aʻo i nā ʻōnaehana aʻo mīkini i ka haʻawina hewa a e hopena i ka hopena hewa. Ua ʻōlelo ʻo ia, "Ua pili ʻia nā ʻikepili āpau. ʻAʻole kēia he paranoia. He ʻoiaʻiʻo kēia. – Kauka Sanjiv M. Narayan, Ke Kula Lapaʻau ʻo Stanford University.

ʻO ka hoʻohana ʻana i ka ʻikepili biased no ka hoʻomaʻamaʻa ʻana ua alakaʻi i kekahi mau hemahema AI kaulana. (Na laʻana maanei a maanei, noiʻi maanei..)

Nā Algorithms Kūlike

ʻO kahi algorithm kahi hoʻonohonoho o nā lula e ʻae i kahi hoʻokomo a hana i nā mea e pane ai i kahi pilikia ʻoihana. He mau kumu hoʻoholo maikaʻi lākou. Ua like nā algorithm me nā pahu ʻeleʻele. ʻAʻohe mea maopopo i ke ʻano o kā lākou hana, ʻaʻole hoʻi i ka nā ʻoihana e hoʻohana ana iā lākou. ʻAe, a he kuleana pinepine lākou. ʻO ko lākou ʻano pohihihi a paʻakikī kekahi o nā kumu i ʻike ʻole ʻia ai nā algorithm biased. . 

E noʻonoʻo i nā algorithms AI i ka lāʻau lapaʻau, HR a i ʻole kālā e noʻonoʻo ai i ka lāhui. Inā he kumu ka lāhui, ʻaʻole hiki i ka algorithm ke makapō lāhui. ʻAʻole kēia he kumu. Ua ʻike ʻia nā pilikia e like me kēia ma ka honua maoli me ka hoʻohana ʻana iā AI in ma waena, ride-share, palapala nois, a hoʻololi ʻana i nā kīkī. 

ʻO ka laina lalo inā he hewa kāu ʻikepili a i ʻole algorithm, ʻoi aku ka maikaʻi ma mua o ka pono ʻole, hiki ke pilikia. Aia kekahi mea e like me "loiloi algorithmic.” ʻO ka pahuhopu ke kōkua i nā hui e ʻike i nā pilikia e pili ana i ka algorithm e pili ana i ka pono, ke ʻano a me ka hoʻokae. Ma kahi ʻē aʻe, Facebook ke hoʻohana nei ʻo AI i ka hakakā ʻana ma AI.

Kūlike nā kānaka

Loaʻa iā mākou nā poʻe ma nā ʻaoʻao ʻelua o ka hoohalike. Ke hoʻomākaukau nei ka poʻe i ka nānā ʻana a ke loaʻa nei i nā kānaka ka ʻike. Aia nā mea noiʻi a aia nā mea heluhelu. Ma kekahi kamaʻilio, hiki ke pilikia i ka hoʻouna ʻana a i ʻole ka hoʻokipa ʻana.

E lawe i ka wā, no ka laʻana. He aha ka manaʻo o "kahi manawa o ka ua"? ʻO ka mea mua, he aha ka manaʻo o ka poʻe meteorologists ke ʻōlelo lākou aia ka manawa o ka ua? Wahi a ke aupuni US Ka Honua Nānā Makahiki, kahi manawa o ka ua, a i ʻole ka mea i kapa ʻia ʻo Probability of Precipitation (PoP), ʻo ia kekahi o nā mea liʻiliʻi loa i hoʻomaopopo ʻia i loko o ka wānana aniau. Loaʻa iā ia kahi wehewehe maʻamau: "ʻO ka probability of precipitation he mea helu helu wale nō o 0.01" ʻīniha [sic] o [sic] ʻoi aʻe o ka ua ma kahi i hāʻawi ʻia i ka wahi wānana i ka manawa i kuhikuhi ʻia." ʻO ka "wahi i hāʻawi ʻia" ʻo ia ka wahi wānana, a i ʻole broadwahi hoolei. ʻO ia hoʻi, pili ka Probability of Precipitation ma luna o ka hilinaʻi e ua ka ua ma kahi o ia wahi a me ka pakeneka o ka wahi e pulu. I nā huaʻōlelo ʻē aʻe, inā hilinaʻi ka mea meteorologist e ua ana ma kahi wānana (Confidence = 100%), a laila hōʻike ka PoP i ka ʻāpana o ka wahi e loaʻa ai ka ua.  

Alanui Parisa; Lā Ua,Gustave Caillebotte (1848-1894) Chicago Art Institute Public Domain

ʻO ka manawa o ka ua ma muli o ka hilinaʻi a me ka ʻāpana. ʻAʻole au i ʻike i kēlā. Manaʻo wau ʻaʻole ʻike nā poʻe ʻē aʻe i kēlā. Ma kahi o 75% o ka heluna kanaka ʻaʻole maopopo i ke ʻano o ka helu ʻana o ka PoP, a i ʻole ka mea i manaʻo ʻia e hōʻike. No laila, ke hoʻopunipuni ʻia nei mākou, a i ʻole, he pilikia kēia o ka ʻike. Kāhea kākou i ka ʻike ua. Hoʻohewa anei mākou i ka mea ʻike wānana? No ka pololei, aia kekahi hilahila ma waena o ka poʻe wānana ka wā. I hoʻokahi anamanaʻo, 43% o nā meteorologists i noiʻi ʻia i ʻōlelo ʻia he liʻiliʻi loa ke kūlike o ka wehewehe ʻana o PoP.

ʻO ka ʻIkepili ponoʻī he Biased

ʻO nā kumu hoʻohālikelike ʻelima, ʻo ka loiloi ponoʻī paha ka mea kupanaha loa. Ma ka noiʻi ʻepekema e paʻi ʻia ana kahi pepa loiloi ʻia, ʻo ka mea maʻamau ke kuhi ʻia ke kumumanaʻo, wehewehe ʻia nā ʻano e hoʻāʻo ai i ke kuhiakau, hōʻiliʻili ʻia ka ʻikepili, a laila nānā ʻia ka ʻikepili. ʻO ke ʻano o ka nānā ʻana i hana ʻia a me ke ʻano o ka hana ʻana ʻaʻole i mahalo ʻia i ka hopena o ka hopena. Ma kahi pepa i paʻi ʻia ma mua o kēia makahiki (Ianuali 2022), ma ka International Journal of Cancer, ua loiloi nā mea kākau inā nā hopena o nā hoʻokolohua hoʻokele randomized a me nā noiʻi nānā ʻana i ka retrospective. Ua hoʻoholo kā lākou ʻike,

Ma ka hoʻololi ʻana i nā koho analytic i ka noiʻi hoʻohālikelike hoʻohālikelike, ua hana mākou i nā hopena kū'ē. Hōʻike kā mākou mau hopena e ʻike paha kekahi mau haʻawina nānā retrospective i kahi lapaʻau e hoʻomaikaʻi ai i nā hopena no nā maʻi, aʻo kekahi noiʻi like ʻole e ʻike ʻia ʻaʻole, pili wale i nā koho analytical.

I ka wā ma mua, i ka heluhelu ʻana i ka puke moʻolelo ʻepekema, inā like ʻoe me aʻu, manaʻo paha ʻoe e pili ana nā hopena a i ʻole nā ​​hopena i ka ʻikepili. I kēia manawa, ʻike ʻia nā hopena, a i ʻole e hōʻoia ʻia a hōʻole ʻia paha ke kuhiakau mua i ke ʻano o ka nānā ʻana.

kekahi like loaʻa nā hopena like. ʻO ka ʻatikala, Nui nā mea loiloi, hoʻokahi pūʻulu ʻikepili: hana akaka pehea e pili ai nā ʻano like ʻole i nā koho analytic i nā hopena, wehewehe i ke ʻano o kā lākou hāʻawi ʻana i ka ʻikepili like i 29 mau hui like ʻole e kālailai. ʻIke pinepine ʻia ka ʻikepili ma ke ʻano he hana koʻikoʻi a wehewehe maikaʻi ʻia e alakaʻi i kahi hopena hoʻokahi.  

ʻOiai ke ʻano o ka hoʻopiʻi ʻana a ka poʻe methodologists, he maʻalahi ke poina i ka ʻoiaʻiʻo e hilinaʻi paha nā hopena i ka hoʻolālā analytic i koho ʻia, kahi i hoʻopiha ʻia me ka manaʻo, manaʻo, a me nā wahi koho. I nā hihia he nui, nui nā ala kūpono (a me nā kumu kūpono ʻole) i ka loiloi ʻana i ka ʻikepili e pili ana i kahi nīnau noiʻi.

Hoʻopili ka poʻe noiʻi i ka ʻikepili o ka ʻikepili a hiki i ka hopena o nā noiʻi āpau e pili ana i nā hoʻoholo kumuhana - me ke ʻano o ka nānā ʻana e hoʻohana ai - hiki ke hoʻopilikia i ka hopena hope loa o ka noiʻi.

ʻO ka ʻōlelo a kekahi mea noiʻi ʻO ka mea nāna i kālailai i ka haʻawina ma luna nei e makaʻala i ka hoʻohana ʻana i hoʻokahi pepa i ka hoʻoholo ʻana a i ʻole ka hoʻoholo ʻana.

Hoʻoponopono i ka Bias ma Analytics

He moʻolelo hoʻomanaʻo wale nō kēia. Hiki i ka ʻike ke pale iā mākou mai ka lawe ʻia ʻana e nā scams. ʻOi aku ka ʻike i nā ʻano hana e hiki ai i ka scanner ke hoʻopunipuni iā ​​​​mākou, ʻoi aku ka liʻiliʻi o ka lawe ʻia ʻana iā mākou, e ʻōlelo, e, e ʻōlelo, ke kuhi hewa ʻana o ka ʻeke ʻeke, a i ʻole ke kamaʻilio mālie o kahi pāʻani Ponzi. Pēlā nō me ka hoʻomaopopo ʻana a me ka ʻike ʻana i nā manaʻo like ʻole e pili ana i kā mākou loiloi. Inā ʻike mākou i nā hopena kūpono, hiki iā mākou ke hōʻike maikaʻi i ka moʻolelo a hoʻoholo i nā hoʻoholo maikaʻi.