I have just uploaded a new version of SwingNN. The crash in pattern forecasting is fixed. All those annoyingly complicated features in pattern detection and ...
Dear Steve The support@... mail does not work Yours Michael ... From: Stephen Wolstenholme To: easynn@yahoogroups.com Sent: Sunday, January 04, 2009...
... Hello, Anything, including the weights, can exported in txt or csv using Tools > Special Files. The nodes and weights can be viewed graphically on View >...
I recently has a conversation with a fellow software writer who, like me, depends of his products for his income. He is a "marketing" type and it shows. The...
Steve. I think you have enough for gloat over., In my case we have used the EasyNN in a lot of project in our Propane Gas Company for years. We never thought...
Cortes Marcelo (GNTE)
mcortes@...
Feb 6, 2009 1:12 pm
1477
Steve, I think if you could find someone who has used the software to win a datamining contest on the international stage that would be extremely useful. I am...
Marc Parham
marc.parham@...
Feb 6, 2009 1:31 pm
1478
Also, Steve you can tell anyone that I use Easynn to do climate and tropical predictions and the later version like 9.0 and later was a great improvement when...
Paul (Skywarn)
wxwatch@...
Feb 6, 2009 11:36 pm
1479
Hi Stephen, I previously owned an older version of EasyNN, and also own NeuroDiet.. You had offered a discount if I bought both EasyNN and the new one...
Hi Myke, My records show that you have EasyNN-plus version 9 and a copy of version 10 on loan. If you go to www.npsl1.com you will see the current offer....
Hello, "Leave Some Out" was added to help one user discover trends and validate a lot of medical data. I rarely get any bug reports or suggestion about the use...
There is a minor update available today. The changes are as follows: 1. A bug in trimming when removing the example with the highest error is fixed. 2. A...
Could I have the URL please. also, I notice a number gave their testimonials via Yahoo whilst I sent mine to you personally, is that ok Steve? I had intended...
YES - is it the only way to really validate a NN model can you set it up to leave out 10% but in manner where you use the same rows in the 10 folds so that the...
Marc Parham
marc.parham@...
Feb 12, 2009 4:40 pm
1485
I will send you the URL. Testimonials have been requested using this group and by sending emails to everyone who registered a product recently. I have received...
... Hi Marc, I think I may be misunderstanding your request. If the same rows and the same random seed are used, every fold will produce the same results. ...
If I may interject here, Steve, if you recall there was a suggestion that I sent in some time ago (couple of years?) suggesting that where a universe of data...
Yes I should have used the word mutually exclusive so that for a 10 fold cross validation each row is only used once in the validation set. and 9 times in the...
Marc Parham
marc.parham@...
Feb 12, 2009 5:56 pm
1490
That's not much of a change to row selection. I'll put it in 10.0i The random seed should not change unless it is enabled in the Forget defaults. Steve...
Okay, to clarify. Does "Leave some out" enable the full set of NN's to be consolidated by e.g. a committee, to produce one overall result for the 10 NN's that ...
can you set the seed to default to the same number (same selection of rows as validate and train) UNLESS the user changes the random seed number ? This way as...
Marc Parham
marc.parham@...
Feb 12, 2009 10:19 pm
1494
Peter, Yes and no. "Leave Some Out" produces a multifold report if the grid has one or more boolean output columns. Tools > Classify Column can be used if...
Marc, The random seed is always the same number unless ReSeed is set in Defaults > Forget. The selection of rows to train or validate is part of setting up...
I was thinking of regression type problems with continuous outcomes. If you could output just the predictions when the rows were in the validate (not used for...
Marc Parham
marc.parham@...
Feb 12, 2009 11:26 pm
1497
Steve, Would it be possible to add feature to EasyNN to do a step forward for regression analysis just like the stepforward process in traditional linear ...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
