5,619 reputation
11035
bio website www54.sap.com/industries/…
location Switzerland
age 38
visits member for 3 years, 11 months
seen 6 hours ago

During the day, I forecast sales at supermarkets, drugstores, furniture, perfume and other retailers and calculate order proposals. Lots of time series, with an emphasis on fast, automatic and robust data cleansing and forecasting - with some logistical optimization thrown in for good measure. I'm active in the International Institute of Forecasters and an Associate Editor for their practitioner-oriented journal Foresight.

At night, I switch hats and do inferential statistics for clinical and biological psychology.

I'm never bored. And I use R.


11h
comment How to find a Threshold Value
So you simply want to cluster your one-dimensional data into exactly two clusters? Check out k-means clustering with k=2.
1d
answered Holt-Winters and Abnormal termination in LNSRCH
1d
answered How to blend multiple time series models?
Aug
10
comment Prediction using Support Vector (SV) method in R
Quite honestly, I wouldn't be using SVM for time series, anyway. I don't even see how to model trend or seasonality using SVM. This may be helpful.
Aug
8
comment How to forecast with quantile regressoin
Forecasting should not be a problem. Cross validation looks a bit more interesting. Could you edit your question to clarify what exactly you are interested in?
Aug
8
comment Prediction using Support Vector (SV) method in R
using SVM method we can predict the future value more accurately than other normal methods (like ARIMA) - certainly not true. This will depend heavily on your specific application. To your question: load the e1071 package with library(e1071), then look at ?predict.svm, especially the examples.
Aug
7
comment confidence interval for coefficient of variation
You could always bootstrap, using the boot package.
Aug
7
comment Applying ARMAX model from r output
Please give a reproducible example. I recommend that you read ?predict.Arima - note that you need to feed prediction external variables into the newxreg parameter.
Aug
7
comment How to build a prediction model for exam score based on previous scores
You say that your $X$ is the date of the test. That does not appear correct - your explanatory variables are the results of past tests, possibly together with the dates of past tests. Is that correct? Perhaps you could edit your question for clarity. As you correctly note, there may be ceiling effects, so it may make sense to convert test scores to the interval [0,1], then do some kind of logistic-type transformation. Data Science may be an alternative venue - consider flagging the question for migration if you don't get an answer here.
Aug
3
comment Does $\Pr(\text{Type I error})$ ever not equal $\alpha$ with continuous data?
You mean apart from misspecification of functional forms, link functions, error distributions, heteroskadastic form and unobserved variables? No... if you have all these right, then mathematics guarantee your alpha levels. (Of course, there are people out there who keep claiming that the null hypothesis does not hold in any question you'd want to examine, anyway. Pay no attention to those people.)
Jul
30
comment Question on unexpected regression results
You recoded your FA as a factor in your lm call. From what you write, FA should be a numerical measurement, shouldn't it?
Jul
30
revised Forecast error differences when using weekly vs monthly data
added tags
Jul
30
answered Forecast error differences when using weekly vs monthly data
Jul
28
revised VIF in GLM model in R
added multicollinearity tag
Jul
28
comment VIF in GLM model in R
A few words on multicollinearity in general are here. The multicollinearity tag may also be helpful.
Jul
28
answered Mcomp rolling forecasts with re-estimation
Jul
25
comment Mcomp rolling forecasts with re-estimation
Stepping though is not the problem. My question is what you want a to do while you are rolling your forecasts. Right now, a is the MAE over the entire holdout sample, but if you do rolling forecasts, a would be based on fewer and fewer holdout observations. Is that what you want?
Jul
25
revised VIF in GLM model in R
edited tags
Jul
25
comment VIF in GLM model in R
Multicollinearity is a property of the regressors, not the model, so you don't need to look for "multicollinearity in GLM" as opposed, say, to "multicollinearity in OLS". In addition, there are other measures of multicollinearity than VIF, like the condition indices and variance decomposition proportions of Belsley, Kuh & Welsch, so it would be good if you could edit your question - are you specifically interested in the VIF, or generally in detecting multicollinearity in R? (I also voted to close and move to stackoverflow.com, since this seems to be specifically about R.)
Jul
21
comment If two time series $X$ and $Z$ follow $0 \leq Z \leq X$, can we say that $\text{var}(Z) \leq \text{var}(X)$?
How does your first "observation" (that $X$ is not constant) follow from your assumptions? I don't see how it does. Indeed, assume that $X=1$ is constant, then of course your conclusion does not follow. Or if you add the assumption that $X$ is nonconstant, then you can have it fluctuate a tiny little bit around 1, as in Alexis' answer, and your conclusion again does not follow.