Hi guido, Just to add to what Andrew said, prediction.io mainly leverages spark’s MLlib and ml libraries, there are templates which use h20 and deeplearning4j also ( and any other java/Scala libs as Andrew mentioned )
Take a look on the website for examples! Note that the templates for some algorithms are older so will require more tweaking to run. On Tue, 27 Feb 2018 at 16:41, Andrew Troemner <[email protected]> wrote: > Hi Guido, > > PredictionIO primarily uses Scala as its numerical processing back-end, so > you should be able to utilize any of the Java standard math libraries or > modify any existing classes as you see fit. It does require compiling new > Scala code as part of your source code, but that's fairly typical with any > customizations or process flows you put in place beyond the default. > > Thanks, > > > *ANDREW TROEMNER*Associate Principal Data Scientist | salesforce.com > Office: 317.832.4404 > Mobile: 317.531.0216 > <http://smart.salesforce.com/sig/atroemner//us_mb_kb/default/link.html> > > On Tue, Feb 27, 2018 at 11:31 AM, Guido Arata <[email protected]> wrote: > >> Hi everyone, I am thinking of using PredictionIO for building my machine >> learning system but I can't seem to find any documents which can tell me >> how to add new, custom, algorithms to the system. My clients for sure will >> need customised algorithms and I would like to understand how flexible is >> Prediction.IO when it comes to implement tailor made algorithms. Anyone has >> experience with this scenario? >> >> Thanks, >> Guido >> >> >> >> >> Guido Arata // CTO >> Skype: guido.arata >> Slack: @g.arata >> Mob ITA: +39.348.6727395 >> Mob SG: +65 9243 2455 <+65%209243%202455> >> >> Via Tacito 6, 20137 Milano - MI >> Tel. +39 0254102393 <+39%200254102393> Fax. +39 025455082 >> <+39%20025455082> >> http://dcg.net >> > >
