Mathematical Statistics (STAT 512) Final

Wow ! What an exam !

Today was the final exam for the Mathematical Statistics course. It was an old fashioned 2 hour in-class exam. I have not given one of these in a while now.  

You know that an exam is going to be quite long when the first lines on the questionnaire read ” This is a long exam and most of you will not be able to complete this in time. But that is alright and will be taken into consideration while grading.”

Indeed it held up to every word of that warning. As soon as I saw the exam sheet, I knew that I needed to step it up a gear for the next couple of hours. That is exactly what I did and went ‘into the zone’ completely. It was the kind of an exam where when you read a problem, the first thing you try for it better be the correct approach, or else you will not finish the exam. 

Somehow I did manage to finish the exam and was quite happy and confident with my attempt. Used 4 answer booklets on the way.

But boy oh boy, haven’t felt such an adrenaline rush since the quals of 2010 🙂

With this exam, come the end of one of the most useful and enjoyable courses I have ever taken. Granted that the course involved a lot of work for a 5th year Phd student but I learnt a lot of useful stuff and from one of the most entertaining and smart instructors. I highly recommend this course.

Thank you Dr Ewens ! 

STAT 501: Non Parametric Statistics and Log-Linear Modeling

Just turned in the take home final exam for Stat 501.

This was a useful course. I have taken a previous course (Stat 500) by Prof Rosenbaum and already like his teaching style.  

The best part of taking this course is that it is quite light-weight. There are no homeworks, only a mid term and a final and both are take-home. 

Just like Stat 500, this course is quite applied, which I really like. I was able to use some of the non-parametric methods just out of the box with my research problem and it resulted in a paper submission. 

Paper marathon week ends !

This has been a super productive week.

I submitted the following 3 papers ! in the past week from March 14-21.

1) IMpACT to CASE 2014

2) Maximum likelihood modeling to CDC 2014, and

3) MLE+ campus wide simulation to IBPSA 2014. 

The cdc and ibpsa submissions were just 12 hours apart so I did not sleep or go home during March 20 and 21st. 

Been a long time since an all nigher but certainly worth the effort. 

Paper Submitted to CASE 2014

I submitted a paper to the special session on simulation and optimization in building automation at the tenth annual IEEE International Conference on Automation Science and Engineering (IEEE CASE 2014)

The paper is an extension to our previous work on uncertainty propagation from sensing to modeling and control for buildings. 

The extensions are two-fold

(1) Unlike the simplified assumption in the ModelIQ paper, here we assume that uncertainty in data is not a fixed bias but is a random gaussian variable .

(2) We use non parametric statistical methods to quantify the uncertainty from real sensor data and present a method to determine the optimal placement and density of the sensors.