NUS SCIENCE(STATISTICS)

[Death4ngel]

maybe........

Maybe? So why do u want to switch discipline?

 

[mgx-alander]

hai, moi math beh kan pls tell moi if i am doing this right for stats

i got P(A) = 0.4, P(B) = 0.2, they are independent so P(A^B) is 0.08

I got a conditional probability of

P(C|A^B) = 0.7
P(C|A^~B) =0.3
P(C|~A^B) = 0.1
P(C|~A^~B) = 0.1

I am asked to find P(C)

Bayes formula put
P(C^A^B) = P(C|A^B) P(A^B)
(i.e. P(A^B) = P(A|B) P(B), let B = (A^B) in this case)



P(A^B^C) = P(C|A^B) P(A^B)
P(A^B^C) = P(C|A^B) P(A|B) P(B) <- cant break C to be alone

the only way to break P(C) alone ish like that
P(A^B^C) = P(A^B|C) = P(A^B) P(C) <- but i dun have P(A^B|C)

so there ish no way i can get P(C) right cos there is always this unknown

 

[mgx-alander]

bump to let moi edited post read

 

[yoshi_fanboy]

I lazy to type out the quantities here, but I will just give you the sketch of the solution in my mind.

Firstly, you know that A and B are independent, so certainly you know that A and ~B, ~A and B, and ~A and ~B are independent. So you can calculate P(A^~B), P(~A^B) and P(~A^~B) given that you know what are P(A) and P(B).

Also, since you know the quantities P(C|A^B), P(C|A^~B), P(C|~A^B) and P(C|~A^~B), by Bayes Rule you can certainly compute P(C^A^B), P(C^A^~B), P(C^~A^B) and P(C^~A^~B) respectively.

Now we know that P(C)=P(C^A^B)+P(C^A^~B)+P(C^~A^B)+P(C^~A^~B), so just use the quantities that you have computed earlier and sum it up to get the desired answer.

 

[mgx-alander]

see i so beh kan

thank chiu

 

[mgx-alander]

Given P(A), P(B) P(C) and P(E) (They are all independent)

I am also given
P(F|D^E), P(F|D^~E), P(F|~D^E), P(F|~D^~E)

I know P(F^D^E) = P(F|D^E) P(D^E) = P(F|D^E) P(D) P(E)


But I have a missing P(D) and i duno P(F^D^E)

I need to find P(D|A^B^C).. hows?

I also need to find P(D|F)

 

[Death4ngel]

What paper is this? Is this a standalone qns?

 

[mgx-alander]

yes this is a standalone qn..

its one of the topic in ai involving bayes theorem.. but they like to give ranjiao qn like this knn


and st2334 isnt a pre req

this paper is old tho.. 2008 s2 nt sure if ppl are taught a technique call variable reduction which i google can solve this qn but im nt taught that 1 so i saw the notes i

but other pyp has bery little qn bout this sub topic so..

 

[Death4ngel]

Got the full qns a not? Like got something missing leh.

 

[Darkzi0n]

cant solve

 

[mgx-alander]

Given P(a) = 0.2, P(b) = 0.3, P(c)= 0.2, P(e) = 0.01, all independent

P(d) is dependent on P(a), P(b) and P(c)

P(f) depends on P(d) and P(e)
the probabilities of P(f|d^e) is 0.95 P(f|d^~e) = 0.8,
P(f| ~d^e) = 0.9, P(f|~d^~e)=0.01 respectively

the only missing piece of information is the conditional probabilities is P(d) for all possible combinations of a, b, and c
i.e P(d|a^b^c), P(d| a^b^~c)... etc

Derive a reasonable estimate for the conditional probabilities of P(d) include ur workings

 

[Death4ngel]

dafuq. reasonable estimate

 

[mgx-alander]

duno wtf the paper huan me to do.. i use bayes theorem convert and convert oso got smth missing seeking sexperts reply

 

[flame_copper]

Given P(A), P(B) P(C) and P(E) (They are all independent)

I am also given
P(F|D^E), P(F|D^~E), P(F|~D^E), P(F|~D^~E)

I know P(F^D^E) = P(F|D^E) P(D^E) = P(F|D^E) P(D) P(E)


But I have a missing P(D) and i duno P(F^D^E)

I need to find P(D|A^B^C).. hows?

I also need to find P(D|F)

I also did something like this in NTU comp sci also,
they taught how to classify iris flower using naive bayes therom,

 

[mgx-alander]

simi iris flower? i nv learn before leh.. think the paper is obselete cos its 5years ago

 

[flame_copper]

simi iris flower? i nv learn before leh.. think the paper is obselete cos its 5years ago

Iris flower data set - Wikipedia, the free encyclopedia

 

[awesomeblue]

Hi guys,
I am a year 1 gg year 2 stats sporean. Anyone here as well?

Notice this thread got a lot nice seniors around like raichu,darkzion and few other.

It is so cool! Since lecture classes are usually filled with ah tiongs i hardly had any friends in stats.

 

[Darkzi0n]

Haha r u the one that emailed me?
Paiseh havnt reply. Jus out pro from ict

 

[mgx-alander]

fwah got new crone

 

[Tatami]

Hello blue! maybe because you are still taking a lot of math mod in year 1, you will be able to make more friends in year 2

Bidding will starts tmr. I'm split between ST4241 (Design and analysis of clinical trials) and ST3246 (statistic model for actuarial science) as my last core mod. I'm not sure if i should take ST4241 as it will be my first year 4 mods and i might be competing with a lot of honour students but i'm more interested in biostatistics. Oh and I haven't take ST3232 Design and Analysis of Experiments... not sure if there is any advantage from taking ST3232 first. Any advice?

Thank you in advance!