869 reputation
16
bio website
location
age
visits member for 1 year, 8 months
seen Aug 28 '13 at 11:45

Jan
8
awarded  Yearling
Aug
16
comment Why do we use a one-tailed test [F-test] in analysis of variance (ANOVA)?
Why are you under the impression that a one-tailed test has to be an F-Test? To answer your question: The F-Test allows to test a hypothesis with more than one linear combination of parameters.
Aug
15
comment What regression method to use?
Ah sorry I just got what you mean. Usually in linear regression you work with an input vector and a parameter vector, giving you one output variable (the data is then in matrix form). What you want are several output variables, right? The consensus on this in another thread was that you should probably do a regression for each individual output variable in the vector, as a combined approach usually does not hold additional information or is mathematically equivalent. Sometimes you have restrictions to take into account. Search for interdependent equation systems, as an example in econometrics.
Aug
12
answered Assumptions of linear regression
Aug
7
comment What is the difference between a hierarchical linear regression and an ordinary least squares (OLS) regression?
I don't think this is completely true, because if you implement dummy variables in the OLS framework, you do not need to assume independence of the locations. In fact you can implement codings for as many different levels and features as you like, though that might be mathematically equivalent to whatever hierachical regression is in the OLS context (I dunno lol)
Aug
7
revised How to prove the correlation between two variables using covariance
added 490 characters in body
Aug
7
answered How to prove the correlation between two variables using covariance
Aug
7
answered Conditional probability with independent events
Aug
6
comment Removing a 'trend in variance' from a time series
Nah it's all good. It always bothers me when I can't learn ALL THE THINGS from all fields anyways. Gonna do a machine learning course eventually as well - and all the "*metrics" have some individuals terminology difference. But of course machine learning really takes the cake in that respect. Really confusing at times :(
Aug
6
comment Removing a 'trend in variance' from a time series
As long as we can both be angry at machine learning for introducing a new terminology we have to learn for the same things ;-)
Aug
6
answered Does talking about a probability of future events make you a Bayesian?
Aug
6
comment Removing a 'trend in variance' from a time series
And because of that focus, many accessible/easy and broad books or information are available for timeseries from the econometrics perspective. And also a lot of money is put in the research of economic or financial data analysis. Maybe I am wrong, but I think finding a physics centered book on time series analysis will probably entail the need for much more theoretical sophistication.
Aug
6
comment Removing a 'trend in variance' from a time series
I meant it the other way around - econometricians don't do more, but the "standard" teaching progression in finance and economics is overwhelmingly focused on large datasets and timeseries data. For better or worse the courses often skip other concepts, more than other fields I think. Also, for me, Hamilton and Luetkepohl are the standard monographic texts in time-series analysis and both are from the economics field. Both Engle and Granger with the nobel price for cointegration and Garch (TS essentials) are economists...
Aug
6
comment Removing a 'trend in variance' from a time series
As far as the general techniques for time series are concerned - (F)GLS is very much a technique associated with cross-sectional yet heteroscedastic data - and data which has explanatory variables. Often in time series you will try to instead model the time series on itself, so with lagged values of that time series as explanatory variables. This is another topic (keywords: Arma, Arima). Once again I strongly suggest looking at econometric literature, as this is the field most concerned with time series. The books from above will be an excellent start.
Aug
6
comment Removing a 'trend in variance' from a time series
well you had the right idea. In general, dividing by the variance gives you a good transformation. This essentially what FGLS does (based on the empirical estimate of your variance). But you should be aware that the precise GLS estimate would be with the unknown, theoretical covariance matrix. Anyway, if you transform via FGLS you should have what you were looking for, good luck.
Aug
5
answered Removing a 'trend in variance' from a time series
Aug
5
revised R squared change multiple linear regression
deleted 18 characters in body
Aug
2
answered R squared change multiple linear regression
Jul
31
comment Conceptual issues in linear modeling
(d) Yes that is correct. w(t), which is our $\epsilon_t / \eta_t$ above, can not be determined for any specific $t$ from the data. It is unobserved. If we knew that we had the absolute true values of our parameters, then we could calculate w(t) / $\eta_t$ which would be equal to the residuals $\hat{\eta_t}$. But there is no way to know if our estimates are the true values. We can only conclude this: If the OLS assumptions hold, then our estimates are correct by expectation/on average and our residuals correspond on average to the error term values - which is why their sum is zero.
Jul
31
comment Conceptual issues in linear modeling
(c) IF the model is correctly specified (among other things ALL necessary variables are included, including what the measurement technician had for breakfast - ie the system is completely deterministic), then OLS will give you the exact, true values of the parameters and the residuals will be zero, correct. But as you can see if you are handling real data, there will always be noise - and this is important because you assume no noise and there is noise, your estimates will be completely wrong!