FAKTOR KEPASTIAN DAN KETIDAKPASTIAN Farah Zakiyah Rahmanti Mei 2015
Overview Penalaran Faktor Ketidakpastian Probabilitas Faktor Kepastian (CF / Certainty Factor)
Penalaran (1) Penambahan fakta-fakta baru membuat tidak konsisten. Karakteristik : Ketidakpastian Adanya perubahan pengetahuan Adanya penambahan fakta baru yang dapat mengubah kesimpulan yang sudah dibentuk
Penalaran (2) Contoh : Premise 1 : aljabar merupakan pelajaran yang sulit Premise 2 : geometri merupakan pelajaran yang sulit Premise 3 : kalkulus merupakan pelajaran yang sulit Kesimpulan : matematika merupakan pelajaran yang sulit Premise 4 : sosiologi merupakan pelajaran yang sulit Tidak berlaku, karena sosiologi bukan merupakan bagian dari matematika Penalaran induktif sangat mungkin adanya ketidakpastian
Faktor Ketidakpastian Kurangnya informasi yang memadai Terhalang untuk membuat keputusan yang baik Salah satu teori yang berkaitan dengan faktor ketidakpastian : probabilitas Bayes
Probabilitas Probabilitas menunjukkan bahwa kemungkinan sesuatu akan terjadi atau tidak. Contoh : Ada 10 alumni Teknik Informatika, 3 diantaranya mahir bahasa pemrograman Java. Maka probabilitasnya p(java) = 3/10 = 0.3
Faktor Kepastian (CF / Certainty Factor) CF menunjukkan ukuran kepastian fakta-fakta atau aturan-aturan. CF [h,e] = MB [h,e] - MD [h,e] CF [h,e] = certainty factor (faktor kepastian) MB [h,e] = measure of belief level to the hypothesis h, if it is influenced by evidence e (between 0 and 1). MD [h,e] = measure of disbelief level to the hypothesis h, if it is influenced by evidence e (between 0 and 1).
Certainty Factor (2) Some evidence are combined to determine the CF of a hypothesis. If e1 and e2 are the observations, then:
Certainty Factor (3) Example : Suppose an observation gives credence to h with MB [h, e1] = 0.3 and MD [h, e1] = 0, then : CF[h,e1]=0.3-0=0.3 if there is a new observation with MB [h, e2] = 0.2 and MD [h, e2] = 0, then: MB[h,e1 ^ e2]=0.3+0.2*(1-0.3)=0.44 MD[h,e1 ^ e2]=0 CF[h,e1 ^ e2]=0.44-0=0.44
Certainty Factor (4) Example : Asih suffers freckles on her face. The doctor estimates that Asih suffered pox with belief MB [pox, freckles] = 0.8 and MD [pox, freckles] = 0.01, then : CF[pox,freckles]=0.8-0.01=0.79 If there is a new observation that Asih is also fever with belief MB[pox,fever]=0.7 and MD[pox,fever]=0.08, then : MB[pox,freckles^fever]=0.8+0.7*(1-0.8)=0.9 MD[pox,freckles^fever]=0.01+0.08*(1-0.01)=0.0892 CF[pox,freckles^fever]=0.94-0.0892=0.8508
Certainty Factor (5) CF dihitung dari beberapa hipotesis. Jika terdapat hipotesis h1 dan h2, maka :
Certainty Factor (6) Contoh : Suppose an observation gives credence to h with MB [h1, e] = 0.5 and MD [h1, e] = 0.2, then : CF[h1,e]=0.5-0.2=0.3 if these observations also give credence to hs2 with MB [h2, e] = 0.8 and MD [h2, e] = 0.1, then: CF [h2, e] = 0.8-0.1 = 0.7 Finding CF [h1 ^ h2, e] is obtained from MB [h1 ^ h2, e] = min (0.5; 0.8) = 0.5 MD [h1 ^ h2, e] = min (0.2; 0.1) = 0.1 CF [h1 ^ h2, e] = 0.5-0.1 = 0,4 Finding CF [h1 h2, e] is obtained from MB [h1 h2, e] = max (0.5; 0.8) = 0.8 MD [h1 h2, e] = max (0.2; 0.1) = 0.2 CF [h1 h2, e] = 0.8-0.2 = 0.6
Certainty Factor (7) Example : Asih suffer freckles on her face. The doctor estimates that Asih suffered pox with belief MB [pox, freckles] = 0.8 and MD [pox, freckles] = 0.01, then : CF[pox,freckles]=0.8-0.01=0.79 If these observations also provide belief that Asih may also affected by allergies with belief MB [allergies, freckles] = 0.4 and MD [pox, freckles] = 0.3 then : CF [allergies, freckles] = 0.4-0.3 = 0.1 Finding CF[pox^ allergies,freckles] is obtained from : MB[pox^ allergies,freckles]=min(0.8;0.4)=0.4 MD[pox^ allergies,freckles]=min(0.01;0.3)=0.01 CF[pox^ allergies,freckles]=0.4-0.01=0.39 Finding CF[pox allergies,freckles] is obtained from : MB[pox allergies,freckles]=max(0.8;0.4)=0.8 MD[pox allergies,freckles]=max(0.01;0.3)=0.3 CF[pox allergies,freckles]=0.8-0.3=0.5
Certainty Factor (7) Conclusion: All belief factors that Asif suffer pox from the appearance of freckles on her face is 0.79. Similarly, the belief factor that Asih suffer allergy from the appearance of freckles on her face is 0.1. with the same symptoms that influence two different hypotheses, so the belief factor become : Asih suffers pox and allergies = 0.39 Asih suffers pox or allergic = 0.5
Reference Ahmad Haidaroh, Pengenalan Kecerdasan buatan, Materi ke-5, STIKOM Artha Buana