Monday, 26 May 2014

Class 5 is now available

The lessons for Class 5, the last in our course, are now available on the course website:

The 6 lessons in Class 5 addresses some important miscellaneous issues. Two are devoted to neural networks, both the simple Perceptron and multilayer Perceptrons — sometimes called “connectionist” models. Then we consider that perennial question, “how much data is enough?”, and show how to answer it using learning curves. Next we look at how to optimise the parameters of learning algorithms, and finally we return to the very beginning and re-visit the ARFF format, including some useful features that haven’t yet been encountered.

The post-course assessment opens on 28th May. Everything will remain open until 11th June, when the course will be closed. The videos, slides and transcripts will remain available at YouTube, Youku and the "Materials" site:

We will also ask you to complete a survey for your opinions of the MOOC.

We are planning to run the introductory course “Data Mining with Weka” again in July. We will run “More Data Mining with Weka” again, but are not sure when: it depends on when we have sufficient new graduates from the introductory course to make it worthwhile. As for a possible third course, “Advanced Data Mining with Weka”, that’s still under consideration: there’s no schedule yet.

Please keep up the help on the course forum -- we greatly value your assistance.

cheers, and enjoy the remainder of the course!


Monday, 19 May 2014

Class 4 is now available

The six lessons for Class 4 are now available on the course website:

In this class we'll learn about two topics: attribute selection and cost-sensitive classification. Automatic selection of an attribute subset is a powerful way of getting both good results and simpler, easily explainable, models from machine learning; indeed you will end up achieving stunning results with a tiny subset of attributes on a document classification task. And taking the costs of different kinds of error into account is essential in many practical applications.

Next week is the last. Pretty soon you will be an expert in data mining and the use of Weka!

It's great to see so many people helping each other on the course forum. Please keep it up -- we greatly value your assistance.

Monday, 12 May 2014

Class 3 is now available

The six lessons for Class 3 are now available on the course website:

After this week there are 2 weeks to go (classes 4 and 5).

It's good to see so many people having fun with the course!

The mid-course assessment is also now available. Do it when you have finished Class 2 (although it will remain open for the rest of the course). The final assessment will appear during week 5.
Occasionally a problem occurs when submitting the assessments, where you click “Submit Answers” and get a blank page. This is a glitch in the Google infrastructure that runs the MOOC, and it is out of our control. Please make a note of your answers before submitting, so that if this happens you can re-submit them without too much pain. Check your Profile for your mark to ensure that it has been recorded.

My goal is to enable you to learn as much as possible from this course, and I recognize that doing the assessments may not be a priority for you. However, our ability to mount follow-up MOOCs will depend on the success of this one as perceived by my University -- and the number of people who complete it successfully will be a key metric. Thus I urge you to do the assessments for my sake, if not your own :-)

cheers, and keep going! Weeks 3 and 4 are the central part of this course.

Monday, 5 May 2014

Class 2 is now available

The six lessons for Class 2 are now available on the course website:

Following that, there are 3 weeks to go (classes 3, 4 and 5).

It's good to see so many people having fun with the course!

The activities are a crucial part of the course: they're where most people will do their actual learning! However, they do not form part of the assessment, so don't be scared to get wrong answers. Also, some of the activities are pretty difficult and time-consuming. You don't necessarily need to actually complete them if you find that difficult on your computer, but you do need to understand what it is that you are supposed to do -- and why.

"More Data Mining with Weka" has been designed so that participants at many different levels can learn as much as possible – and complete the course successfully. You don't have to do the reading. All you must do to succeed are the mid-course and final assessments -- which you can try as often as you like. The mid-course assessment will become available this Friday (May 9) and remain open for the rest of the course. The final assessment will appear during week 5.

The videos and other course components for Classes 1 and 2 can be downloaded from the "Materials" site, in case you find that more convenient than viewing them online:
We plan to put out a short (optional) update video soon.

cheers, and keep going!