So... what did everyone think? I don't think I did too great The last question (19 marks) was easy, but the NCD question was weird and quite a few of the shorter questions confused me so I couldn't finish them
NCD question was wierd, and I wasted lots of time on it. However I think I got it roughly right. I think you needed to use the Chi-squared distribution, we was a bit odd. The last question was nice, but I didn't finish the bit in the middle, where you had to derive the (100+M)e^-M/50 bit. I know I made mistakes e.g. on scaled deviance, but I think I did well enough on everything to pass. Annoying that there wasn't really anything on Timeseries, as I spent a lot of effort studying them. Wishing you all well on the results
Well I think I did question 1 right...but thats about it! Seriously though I thought the NCD question was tough and spent so long trying to figure it out I ran out of time for some of the other parts. I thought it was slughtly harder than the past papers I tried. I'll be amazed if I dont have to re sit this one
Agree the NCD question was odd. The sort of question where I can almost guarantee I interpreted the givens wrong in some way. Apart from that the paper had rather more theory than I was expecting. Was okay. Hopefully that's exorcised it so I can start thinking about something else. Fingers crossed for us all...
Chi-Squared distribution!!!! Why? How? Im thinkin I did worse on this paper than I originally thought. There were lots of 'non-standard' questions on the paper again, same as the April sitting (makes sense I guess as it is the same author - I think). Anyway. Good luck folks.
My thinking is that the sum of two exponential distributions is a Gamma(2,l) distribution, but that the CDF doesn't exist for this, so you need to transform it to a Chi-Squared distribution. Alternatively you could produce the convolution (Which is probably what they wanted you to do), however this would just result in the Gamma distribution anyway... However I may well be wrong...
All sounds far too confusing...what happened to the relatively straight-forward marks from proving a series was stationary, and then giving you 5 marks for calculating autocollerations / autocovariances!
I know the genaral feeling from everybody coming out of my exam hall was holy cr*p, what the hell was that !! So I have resigned myself to the fact I will be repeating it in April. I only hope that this "Author" has decided to take next year off