I recently submitted a pull request to build HTM Studio on Linux. This PR is being finalized. In the meantime feel free to grab and install HTM Studio packages for Ubuntu 14.04 and 16.04 from the URL below.
While installing a package you see some errors about other packages missing or outdated, make sure your Ubuntu distro is up to date (sudo apt-get upgrade)
Download link : https://www.dropbox.com/sh/fgnrxadt07jg9pp/AAAQVY-vIYazhLYFx03Osczra?dl=0
Let me know if this works for you. Thanks !
Disclaimer Please note that those packages are NOT officially endorsed by Numenta. They are provided as is without any warranty of any kind.
Thanks for all the hard work @laurent, this is great!
Just a quick update to let you kinow that I had to move the Ubuntu packages for HTMStudio to a new place . The link to follow is now :
As always, please note that these packages are not officially endorsed by Numenta.
I wonder if you (or anyone) could direct me to the parameter values used in HTM Studio for the NYC Taxi data?
I’d love to test as close to 1:1 against my own attempted implementation as possible. I know it probably using RDSE, though I’m also using my own simple scalar encoder so any recommendations for ‘w’/‘n’/‘min’/‘max’/‘radius’/‘resolution’ would be most welcome.
I looked at the histogram of the rider values to get a sense for this, and fitting that nyc cab data seems to be flipping you off.
HTM Studio uses this function to create all models, just like NAB.