Missing time-series vs. Empty time-series

Lokad is about time-series forecasting, but as simple as the time-series model may seem to be (after all a time-series is nothing more than a list of time-value pairs), there are several subtleties in the way to manage time-series. In this post, we will see how the Lokad time-series model distinguishes missing time-value pairs from empty time-value pairs. Since the topic is slightly complex, I would suggest, if you’re not familiar the Lokad technology, to have a look at our User Guide (in particular, the Forecasting tasks section).

More on WS directories - BindingPoint.com is over

In my previous post, I was reporting that BindingPoint.com was badly dysfunctional. Well, the problem has been solved, BindingPoint.com is no more. On their home page, they blame the market for being too slow to adopt Web Services. Well, I do agree that surprisingly the adoption of web services has been fairly slow; yet, you can’t blame the market for obvious bugs in your web application. Also, Lokad.com (my company, which provides time-series forecasting web services) has been listed in eSigma.

What do most WS directories have in common?

Most Web Services directories have one thing in common: they are totally bugged at the point of being totally unusable. Indeed I have tried to submit the Lokad Forecasting Web Services to several directories. Namely: BindingPoint.com: registration process crashes and the ASP.Net default exception page. WSIndex.org: can’t even login, gets a fatal cgi-bin error while trying. XMethod.com: website painfully designed, registration succeeds but submission crashes. Dmoz.org: “Submit URL” gets me to a page Service Temporarily Unavailable (it has been that way for the last 2 weeks)

Do not make a sum with your forecasts - a coconuts study

Although we have tried to make Lokad as simple and intuitive as possible, statistical forecasting is a counter-intuitive science with many traps. In this post, I am going to describe one of the most frequent mistakes that I have encountered within many companies. In a nutshell, it is wrong to make a sum of forecasted values. Since the problem is quite hard to grasp, let’s start with an example. Let’s say that you have 3 shops; and that those 3 shops are selling coconuts.

Lokad.com, data mining and grid computing

I have finally released my latest project named Lokad.com. Lokad is about time-series forecasting, I believe that our approach is quite radically different from what the market was offering until now. Since Lokad has its own blog, my two posts about the Lokad release can be found here and here.

A few tips for source code versioning (do not drive your co-workers mad)

Source control management (SCM) is a technical matter as well as a good practice matter. Here is a small list of tips that I have found quite useful in practice. A good commit is like a good paper: It starts with an evocative title. Ok, there is no title in SCM but there are comments provided while committing. If your SCM comment is not clear, then how do you expect your co-workers to keep track of what you are doing?

A small developer-oriented PowerShell wish list

I have just started to use the PowerShell a few days ago; and I am more and more impressed by the work that have been done by the MS folks. Yet, being a developer, I have the feeling that many aspects of PowerShell still need to be polished. Too bad that there is no Visual Studio project templates for CmdLets. Providing an hello-world CmdLet with its associated SnapIn would really make the life of the developers easier, smoothing the learning curve.

Let's get cosmetic on C# documentation

I am gathering here just a few C# documentation cosmetic tips that I apply in my own projects. I do not pretend to have achieved any absolute truth or any optimal practice with those tips. I just find them convenient when applied with consistency. #region #endregion directives can be great, but caution not to overuse them by putting a dozen of directives within a 200-lines long code file. Indeed, if region makes the code structure more apparent, region also makes the code less proofreadable because you have to open the regions to actually start reading.

Blogosphere quatitative elements, a few slides

Being part of the Corps des Telecoms, I have been assigned for a modest blog study within the scope of a communication course. I have found the subject highly interesting because blogs open whole new directions of research and whole new methodologies. Indeed, contrary to most usual communication events (ex: person-to-person talks), it’s potentially possible to retrieve most of the blogosphere content through web crawling intensive methods. The slides of the talk have been uploaded.

ResxEditor reloaded - version 1.2 released

They were a couple of long standing issues with ResxEditor. Most of them were actually reported as comments on the blog post of the initial ResxEditor release. All of those issues are now fixed. Bug fixes and new features are detailed on the ResxEditor page. Special thanks to Nick Pasko for carrying most of the work and finding a solution to get rid of the previous cell saving behavior that was driving translators nuts.