Questions on my online forecasting course

I’ve been getting emails asking questions about my upcoming course on Forecasting using R. Here are some answers.

Do I need to use the Revolution Enterprise version of R, or can I use open-source R?

Open source R is fine. Revolution Analytics is organizing the course, but there is no requirement to use their software. I will be using open source R with Rstudio for demonstrating things in lectures.

Is the course free?

No. It costs $1000 to register for the course. The course is being run jointly by Revolution Analytics and Monash University. Most of the money will go to Monash University, and will fund a research assistant to help me with R package development. My R packages and my FPP textbook are free, but they do take time and resources to produce. This course is one way I am funding them.

Can I watch the lectures in my own time?

Provided you are registered, you will be able to log in and watch the lectures after the event. If you really want to, you can watch them repeatedly. This is particularly important for people in Eastern Europe or West Asia where the lectures are in the middle of the night.

Will I be able to ask questions?

Yes. The lectures are meant to be interactive, for those watching live. You can ask questions, interject, etc. I will also try to respond to questions after lectures, especially for those who cannot watch live.

If I’ve already read your books and blog, will I learn anything?

Probably. The course is based on my FPP book, so don’t expect me to cover things that aren’t discussed there. But most people find that they learn more through asking questions, seeing examples worked out, discussion, etc.

I have very little experience with R, short of playing around with Rcmdr. Is that ok?

It is worth brushing up on R if you don’t use it regularly. I will not be using Rcmdr. Instead, I will use Rstudio and we will be learning the relevant commands. A suitable intro is the tutorial at

If you feel that is not enough, try working through the tutorials at

Is there a reading list I can start on leading up to the lectures?

The course is not difficult from a mathematical/statistical perspective, and as long as you are not freaked out by a few equations you should be fine. If you wanted to brush up, go over multiple regression at an introductory level. A suitable background on this is chapter 5 of FPP, or chapter 3 of Pardoe’s Applied regression modeling. But you may have something else handy that would be equally suitable.

What else can I do to prepare?

If you really want to get a head start, try working through Michael Lundholm’s tutorial on R’s time series facilities.

I want to learn about forecasting xxxx. Is this course useful for this purpose?

The course is about time series forecasting. That is, when you want to forecast data collected regularly over time. So if you have annual profits, or monthly sales data, or weekly electricity demand, or daily passenger numbers, or something else collected regularly over time, then this course should be helpful. But if your forecasting problem does not fit into that paradigm, then it is probably not the course you need.

Related Posts:

  • Pingback: Hyndsight - Online course on forecasting using R()

  • Raffael Vogler

    The two essential questions for me to decide whether this course is worth $1000 is:

    1) Are there going to be exercises and will those be corrected by your staff?
    2) Can you earn a certificate?

    I am currently following a couple of courses about ML and statistics offered by Standford and Caltech – the lectures are AWESOME, the exercises are great and make sense und in some cases you can earn a certificate by getting enough points in exercises and final exam.

    And it costs – $0. I know this for-free-attitude is problematic – I am willing to pay money on service – but $1000 is a lot of money and there has to be a good reason to pay that – especially considering the market situation.

    To make a long story short – I would love to follow this lecture in near future on or!

    Kind regards


    • 1. Yes there will be exercises. No they won’t be marked by me or my staff. We will discuss the solutions in the online classes. I will probably post solutions to class members.

      2. Monash Uni will not provide a certificate. I suspect Revolution Analytics would be happy to provide a certificate of participation. I’ll ask them.

      I cannot use edX or Coursera because they only take courses from university affiliated with them. My university has just joined the Future Learn platform, so it is possible that the course will be offered there one day, but there are no current plans to do so.

      In principle, anything free is being sponsored by someone. My textbook is sponsored by, and my R packages are sponsored by me and CRAN. Unless someone is willing to sponsor an online course on forecasting, it is going to cost something.

      Unfortunately, I don’t know of any free online courses on forecasting that I can recommend.

      • Thanks for suggesting the Future Learn platform and OTexts. I will have a close look at both – especially your book on forecasting!

      • ilyakipnis

        Well, on Coursera, Eric Zivot has an intro/review course on econometrics, which includes some time series, and it’s all in R.

        • I didn’t know that. I’m sure Eric will give a great course. But I doubt he will say much about forecasting.

  • $1000 is not expensive for a course taught by an internationally recognized professor. However, if I were running Revolution Analytics’ marketing department, I would cover all the expenses you need and make this course for free to a broad audience.

  • Jason

    Will this course be offered again in the near future?

  • Arsa Nikzad

    Is this course videos available anywhere to purchase?

    • No. But I will be doing a datacamp course on forecasting, sometime next year.

  • diego

    watching the slides for this course and reading the chapter on dynamic regression in your fpp ebook, I have a question on differencing dummy variables when applying dynamic regression. In your ebook you say that the first step is> “1.Check that the forecast variable and all predictors are stationary. If not, apply differencing until all variables are stationary. Where appropriate, use the same differencing for all variables to preserve interpretability.”

    Is it “appropiate” to difference dummy variables when differencing the dependent variable?

    PS: my dummy variables are for weekdays

    • Yes.

      • diego

        But since my dummy is binary what would be the interpretation of the negative 1s that arise when differencing? Fr my point of view it would penalize ‘Tuesdays’ for not being ‘Mondays’ by the same effect