Hyndsight

 

A blog by Rob J Hyndman 

Twitter Gplus RSS

Posts Tagged ‘forecasting’:


IJF quality indicators

Published on 17 May 2013

I often receive email ask­ing about IJF qual­ity indi­ca­tors. Here is one I received today. Dear Pro­fes­sor Hyn­d­man, I recently had a paper pub­lished in IJF enti­tled, “xxxxxxxxxxxx”. I am very pleased with the pub­li­ca­tion and con­sider IJF to be an excel­lent out­let for my work in time-​​​​series econo­met­rics. I have an unusual request, but I hope you will con­sider respond­ing. My research is judged by non-​​​​economists and IJF is not on their list of “qual­ity” jour­nals. It makes a sig­nif­i­cant dif­fer­ence in my research rat­ing and pay. Would you mind send­ing some objec­tive infor­ma­tion re the qual­ity of IJF that I can pass along to the com­mit­tee? And here is part of my reply: The IJF is ranked A in Aus­tralia (we have four lev­els — A*, A, B and C).† The IJF 2011 2-​​​​year impact fac­tor is 1.485. In 2010 it was 1.863. The five year impact fac­tor is 2.450. Com­pare this to the Jour­nal of Busi­ness and Eco­nomic Sta­tis­tics which has a 2-​​​​year impact fac­tor of 1.693, or Com­pu­ta­tional Sta­tis­tics & Data Analy­sis with 1.089. We are ranked 40 out of 305 eco­nom­ics jour­nals based on our 2-​​​​year impact fac­tor. We receive about 400 sub­mis­sions annu­ally, and pub­lish about 70 per year. But that includes invited papers. Of the

(More)…

 
No Comments  comments 

Forecasting annual totals from monthly data

Published on 15 May 2013

This ques­tion was posed on cross​val​i​dated​.com: I have a monthly time series (for 2009–2012 non-​​​​stationary, with sea­son­al­ity). I can use ARIMA (or ETS) to obtain point and inter­val fore­casts for each month of 2013, but I am inter­ested in fore­cast­ing the total for the whole year, includ­ing pre­dic­tion inter­vals. Is there an easy way in R to obtain inter­val fore­casts for the total for 2013? I’ve come across this prob­lem before in my con­sult­ing work, although I don’t think I’ve ever pub­lished my solu­tion. So here it is.

 
2 Comments  comments 

My new forecasting book is finally finished

Published on 21 April 2013

My new online fore­cast­ing book (writ­ten with George Athana­sopou­los) is now com­pleted. I pre­vi­ously described it on this blog nearly a year ago. In real­ity, an online book is never com­plete, and we plan to con­tin­u­ally update it. But it is now at the point where it is suit­able for course work, and con­tains exer­cises and ref­er­ences. We hope that users (espe­cially other lec­tur­ers) will sub­mit mate­ri­als such as slides and exer­cises, that can be shared on the web­site. For those want­ing a print ver­sion, we will be sell­ing it via Ama­zon in the next few months. The online ver­sion will remain freely avail­able. If other authors are inter­ested in this pub­lish­ing model, please see this page. The book is being pub­lished by OTexts, a new inno­v­a­tive pub­lish­ing com­pany I am estab­lish­ing. The fore­cast­ing book is our first pub­li­ca­tion, but we have three oth­ers that should be online within the next month or two. 

 
No Comments  comments 

George E P Box (1919−2013)

Published on 31 March 2013

Last Thurs­day (28 March 2013), George Box passed away at the age of 93. He was one of the great sta­tis­ti­cians of the last 100 years, and leaves an aston­ish­ingly diverse legacy. When I teach fore­cast­ing to my sec­ond year com­merce stu­dents, we cover Box-​​​​Cox trans­for­ma­tions, Box-​​​​Pierce and Ljung-​​​​Box tests, and Box-​​​​Jenkins mod­el­ling, and my stu­dents won­der if it is the same Box in all cases. It is. And we don’t even go near his work on response sur­face mod­el­ling, design of exper­i­ments, qual­ity con­trol or ran­dom num­ber gen­er­a­tion. Occa­sion­ally, a stu­dent won­ders if box­plots are also due to GEP Box, but they were the brain­child of his good friend John W Tukey. I often quote Box’s famous words to my stu­dents “All mod­els are wrong but some are use­ful” (Box, GEP, 1979, Robust­ness in the strat­egy of sci­en­tific model build­ing, Robust­ness in Sta­tis­tics, Aca­d­e­mic Press, pp.201–236.) This sum­marises my view of sta­tis­ti­cal mod­el­ling per­fectly — no-​​​​one should believe their mod­els; instead, treat them as tools to be used to assist in under­stand­ing the data.

 
4 Comments  comments 

The difference between prediction intervals and confidence intervals

Published on 13 March 2013

Pre­dic­tion inter­vals and con­fi­dence inter­vals are not the same thing. Unfor­tu­nately the terms are often con­fused, and I am often fre­quently cor­rect­ing the error in stu­dents’ papers and arti­cles I am review­ing or edit­ing. A pre­dic­tion inter­val is an inter­val asso­ci­ated with a ran­dom vari­able yet to be observed, with a spec­i­fied prob­a­bil­ity of the ran­dom vari­able lying within the inter­val. For exam­ple, I might give an 80% inter­val for the fore­cast of GDP in 2014. The actual GDP in 2014 should lie within the inter­val with prob­a­bil­ity 0.8. Pre­dic­tion inter­vals can arise in Bayesian or fre­quen­tist sta­tis­tics. A con­fi­dence inter­val is an inter­val asso­ci­ated with a para­me­ter and is a fre­quen­tist con­cept. The para­me­ter is assumed to be non-​​​​random but unknown, and the con­fi­dence inter­val is com­puted from data. Because the data are ran­dom, the inter­val is ran­dom. A 95% con­fi­dence inter­val will con­tain the true para­me­ter with prob­a­bil­ity 0.95. That is, with a large num­ber of repeated sam­ples, 95% of the inter­vals would con­tain the true parameter.

 
18 Comments  comments 

ETS models now in EViews 8

Published on 1 March 2013

The ETS mod­el­ling frame­work devel­oped in my 2002 IJF paper (with Koehler, Sny­der and Grose), and in my 2008 Springer book (with Koehler, Ord and Sny­der), is now avail­able in EViews 8. I had no idea they were even work­ing on it, so it was quite a sur­prise to be told that EViews now includes ETS models.

 
No Comments  comments 

Out-​​of-​​sample one-​​step forecasts

Published on 14 February 2013

It is com­mon to fit a model using train­ing data, and then to eval­u­ate its per­for­mance on a test data set. When the data are time series, it is use­ful to com­pute one-​​​​step fore­casts on the test data. For some rea­son, this is much more com­monly done by peo­ple trained in machine learn­ing rather than sta­tis­tics. If you are using the fore­cast pack­age in R, it is eas­ily done with ETS and ARIMA mod­els. For exam­ple: library(forecast) fit <- ets(trainingdata) fit2 <- ets(testdata, model=fit) onestep <- fitted(fit2) Note that the sec­ond call to ets does not involve the model being re-​​​​estimated. Instead, the model obtained in the first call is applied to the test data in the sec­ond call. This works because fit­ted val­ues are one-​​​​step fore­casts in a time series model. The same process works for ARIMA mod­els when ets is replaced by Arima or auto.arima. Note that it does not work with the arima func­tion from the stats pack­age. One of the rea­sons I wrote Arima (in the fore­cast pack­age) is to allow this sort of thing to be done.

 
11 Comments  comments 

Statistical consulting in Australia

Published on 11 February 2013

There must be dozens of sta­tis­ti­cal con­sult­ing busi­nesses and orga­ni­za­tions in Aus­tralia, each spe­cial­iz­ing in dif­fer­ent areas. I do some con­sult­ing work myself, mostly in the fore­cast­ing area, but some­times in other areas of applied sta­tis­tics includ­ing expert wit­ness work in court cases. Email me if you have a project you would like me to take on. How­ever, I often refer poten­tial clients to other sta­tis­ti­cal con­sult­ing groups, as I only have a lim­ited amount of time I can spend on con­sult­ing projects.

 
1 Comment  comments 

Batch forecasting in R

Published on 7 January 2013

I some­times get asked about fore­cast­ing many time series auto­mat­i­cally. Here is a recent email, for exam­ple: I have looked but can­not find any info on gen­er­at­ing fore­casts on mul­ti­ple data sets in sequence. I have been using analy­sis ser­vices for sql server to gen­er­ate fit­ted time series but it is too much of a black box (or I don’t know enough to tweak/​​manage the inputs). In short, what pack­age should I research that will allow me to load data, gen­er­ate a fore­cast (pre­sum­ably best fit), export the fore­cast then repeat for a few thou­sand items. I have read that R does not like ‘loops’ but not sure if the cur­rent cpu power off­sets that or not. Any guid­ance would be greatly appre­ci­ated. Thank you!!

 
8 Comments  comments 

forecast package v4.0

Published on 3 December 2012

A few days ago I released ver­sion 4.0 of the fore­cast pack­age for R. There were quite a few changes and new fea­tures, so I thought it deserved a new ver­sion num­ber. I keep a list of changes in the Changelog for the pack­age, but I doubt that many peo­ple look at it. So for the record, here are the most impor­tant changes to the fore­cast pack­age made since v3.0 was released.

 
22 Comments  comments 
« Older Entries