Peter Rossler
Having a 20+ year procurement background, intrigued by the possibilities that data science and process mining offer, I would love to move into this "new" field :).
Location Barcelona (Spain) and Ludwigsburg (Germany)
Activity
-
Peter Rossler made a comment
Generally, it was fine (I guess given that I had completed the first course only 2 weeks ago ;). Did not remember how to use the AddExpression filter, however, after a quick google-check, I was reminded how to use it (using terms like a1, a2 instead of naming the attributes length or width).
Very good exercise and great way to refresh what was taught before :). -
Peter Rossler made a comment
Happy to continue the previous course (Data Mining with Weka) with you Ian. I am sure that it will be equally fun, well explained and beneficial.
Also happy to see that the plant behind you is growing as well ;). -
Peter Rossler replied to Peter Rossler
Happy to hear that plant and hair are real :). Greetings to beautiful New Zealand.
-
Peter Rossler made a comment
Ian, the plant behind you does not seem to change from week to week (in contrast to your hair ;). It is indeed a nice one and I assume it is artificial (as you like music but might not have a "green thumb"?!?)
Apart from this distraction, the course is very instructive and fun. Thank you :). -
Peter Rossler made a comment
Really great course, previous to the course I had problems well understanding the NaiveBayes Algorithm, now it is clear! Thanks :)
-
Peter Rossler made a comment
Well explained concepts, thank you! It is fun doing the course!
Having studied classification, decision trees, etc. theoretically, it is great to get the concepts explained once more in this way, more hands-on, applying them in the Weka tool. -
Peter Rossler made a comment
Taking Stephanie Williams' advice, here is what I told my wife and my rubber-duck taking my last bath :) :
>use a baseline classifier (like ZeroR => confidence = %-age of class with most occurrences) = minimum or baseline of confidence factor,
>decide on repeated holdout or cross-validation (depending on the number of itemsets available), to get to a... -
Peter Rossler made a comment
91.6% correctly classified instances however choosing at >10 rectangles (accordingly not very efficient, I guess ;)
-
Peter Rossler made a comment
Hi, thanks for the course :). I am Peter, German, 50 years old, self-employed (indirect procurement consulting), and am fascinated by the possibilities of analyzing, interpreting etc. data. I so far completed courses on data mining, process mining and want to add the ability to make use also of Waka and apply the learned to real data and cases.
I am looking...