So, throughout the course of Applied Financial Engineering, Antonio stressed the importance of brute force methodology, but on the last day of the classes today [class before exam], he threw some light on what-if the computing power available is not enough to apply brute force. What if we cannot generate 100,000 random numbers for our calculations, what if we have to work with a limited set of only 1000 past values?
Well, the only option available is then to get the values under constrained samples and reduce the error margin as much as possible. So, it was today that we saw various techniques to reduce the variance of calculations, but considering the processing time as an additional attribute to go along. Eventually, Antonio shared with us a research done by one of the PhD students which clearly showed the amount of variance reduction each method discussed in class can bring in the results for different number of samples, and also the amount of time taken by each method for these samples.
And to bring a sense to all of this discussion, Antonio aptly told us to consider a product of both processing time and the variance reduction value, as our criteria for evaluating each of these methods.
Anyways, spent entire afternoon today to work on the Internet as a Sales Channel group work and now, I also need to complete the Internet as a Sales Channel individual writeup that I have started working on yesterday, and start preparing for exams today. Its going to be long nights these couple of days, and I am all set to live every bit of them.
Well, the only option available is then to get the values under constrained samples and reduce the error margin as much as possible. So, it was today that we saw various techniques to reduce the variance of calculations, but considering the processing time as an additional attribute to go along. Eventually, Antonio shared with us a research done by one of the PhD students which clearly showed the amount of variance reduction each method discussed in class can bring in the results for different number of samples, and also the amount of time taken by each method for these samples.
And to bring a sense to all of this discussion, Antonio aptly told us to consider a product of both processing time and the variance reduction value, as our criteria for evaluating each of these methods.
Anyways, spent entire afternoon today to work on the Internet as a Sales Channel group work and now, I also need to complete the Internet as a Sales Channel individual writeup that I have started working on yesterday, and start preparing for exams today. Its going to be long nights these couple of days, and I am all set to live every bit of them.
No comments:
Post a Comment