Messages
In the seminar I falsely assumed distributions for Xi's individually while the model tells us only how the sum behaves.
In 3b we should use the CI we found in 2c)
lambda hat +- z(a/2)*sqrt(lambda hat/t) ,
and insert for the point estimate of lambda hat and t=36.
We get a smaller CI [679,697] compared to that found in 3a based on the normal distribution, [652,724], as the Poisson standard error is small when t is large. The point estimate is the same, only the error structure is different.
Special thanks to Margarita for pointing out this mistake.
The Department of Economics is currently reviewing the study programs, with the aim of improving the programs. In this process we would greatly benefit from feedback from the students.
We would therefore ask you to spend a short amount of time to respond to questions about the courses given in 2016.
Link to form: https://nettskjema.no/a/economic-master.html
Suggested solutions (S2.3.4) hang on my door.
Make appointment by e-mail.
-Y
The regular exam 2012H on the Laplace distribution can be solved using course material covered until now (with exception of section A.i of Problem 2only). Try solving the questions before you look at the "sensorveiledning". Section A.ii (Problem 2)contains an example of constructing a confidence interval using Slutsky's lemma. Further "sensorveiledninger" will be put out on the webpage later.
mgf in suggested solution used a different parameterization for gamma distribution than what Rice would suggest.
Interested students can look here for two common parameterizations for the gamma distribution
Learning - Wikipedia
https://en.wikipedia.org/wiki/Relationships_among_probability_distributions
https://en.wikipedia.org/wiki/Gamma_function
https://en.wikipedia.org/wiki/Unbiased_estimation_of_standard_deviation
https://en.wikipedia.org/wiki/Bessel%27s_correction
https://en.wikipedia.org/wiki/Consistent_estimator
https://en.wikipedia.org/wiki/Moment-generating_function#Calculations_of_moments...
The correct e-mail address is yudi.wen@econ.uio.no