I want to deploy my Rapidminer model to End User Application

inceptorfullinceptorfull Member Posts: 44 Contributor II
edited November 2018 in Help
dear all, first I have no idea about programming nor deploying

I created a model so data set entered then processed using stacking including NN and K-nn and get the results, so how to deploy that model to make it end user application so can enter NEW data and get predication? so what should I do as new starter to make such application?
Tagged:

Answers

  • Marco_BoeckMarco_Boeck Administrator, Moderator, Employee, Member, University Professor Posts: 1,993 RM Engineering
    Hi,

    sounds like a job for a web service on RM Server :)
    You can pass in data in various formats, e.g. JSON or even Excel files and get the prediction outcome in various formats.

    Regards,
    Marco
  • JEdwardJEdward RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 578 Unicorn
    Yes, particularly if you are not experienced with programming then the webservices creation with RapidMiner is very straightforward. 
    There's some useful videos on Youtube describing how to create one from a process you have made. 
  • inceptorfullinceptorfull Member Posts: 44 Contributor II
    Thanks all can i deploy it on pc application or have to be on web server
  • Marco_BoeckMarco_Boeck Administrator, Moderator, Employee, Member, University Professor Posts: 1,993 RM Engineering
    Hi,

    you need the Server for that.

    Regards,
    Marco
  • inceptorfullinceptorfull Member Posts: 44 Contributor II
    thanks a lot i got it and will try, if have any tutorials for deploying it into Standalone application will apperciate it so I can make it work without the server later on
  • StaryVenaStaryVena Member Posts: 126 Contributor II
    Hello,
    I'm not sure, but maybe this post is what you want:
    http://rapid-i.com/rapidforum/index.php/topic,5807.0.html
    (I'm not sure if it works for RM7.)

    Best wishes,
    Vaclav
  • inceptorfullinceptorfull Member Posts: 44 Contributor II
    thanks for that, I found how to run it with java, but I couldnot understand . also does it work on new dataset?

    is there a way to import the PMML model then deploy it in standalone java application or any coded application so insert new dataset and get results?
  • StaryVenaStaryVena Member Posts: 126 Contributor II
    Hello,
    if you have created process (with port connected  to the process input) and  new Example set, you can process this exampleSet by calling:
    process.run(new IOContainer(exampleSet))
    Vaclav
  • inceptorfullinceptorfull Member Posts: 44 Contributor II
    thanks all, I just want to get the java classes of my process so I can deploy in java application to be used to predict new dataset
  • StaryVenaStaryVena Member Posts: 126 Contributor II
    There is no java classes of process. Process is represented only by xml which is deserialized to objects in memory.
  • JEdwardJEdward RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 578 Unicorn
    Here's a quick explanation on how it works.

    You build RapidMiner processes using the GUI which stores the workflow as an XML file. 
    On running a process in RapidMiner the XML is read by the RapidMiner Engine and it follows it as a set of instructions.

    For example
    • Step 1: Read Data
    • Step 2: Train Naive Bayes Model
    • Step 3: Read NewData
    • Step 4: Apply Model to Data
    • Step 5: Store Results
    The RapidMiner engine might call many different Java classes to do this.  It's extremely difficult and impractical to separate each class out for the specific process you are running. 

    So, the solution is to include RapidMiner as the engine within your java program. 
    In your Java program you then ask the RapidMiner engine to run the process XML you have created. 
    This has the advantage that when you need to update your model you just replace the XML with a new one you have created with RapidMiner Studio.

    Hope that makes it clearer.

    You can get an OEM licence to do this from RM.  Play with RM 5.3 to do it in earlier versions.
    I believe the Server licencing also covers it too so you can have the option of WebService or Java calls. 
  • inceptorfullinceptorfull Member Posts: 44 Contributor II
    really thank you for your helpful and detailed answer, but seems it need a programmer to do so, also it will make me still need rapidminer to get the XML of the new data.

    thanks a lot
  • JEdwardJEdward RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 578 Unicorn
    You don't have a programmer for your existing application?  You might need to explain a little.  I'm sure there's several people on this forum who can offer assistance integrating into your application for consulting costs so you might not need to hire a programmer. 

    Also the reason Marco & I suggested RapidMiner Server initially is that you don't need to be a programmer to create a deployable process.  It's really easy to create WebApps & Web Services which you can call from other applications just by using a URL. 
    Your users can login to a web interface, upload their data and receive the predictions to download.  Really easy to do and no programming required.
  • inceptorfullinceptorfull Member Posts: 44 Contributor II
    I applied for Rapidminer server Academic licence but not accepted I donot know why, I wish I can try it , guess it is my last solution , if no one give me a guide for what to learn to create that application.

    what should I do with macro? and thanks for your follow up
  • inceptorfullinceptorfull Member Posts: 44 Contributor II
    finally got rapidminer server license, so can anyone provide me with tutorial so I can make my predicitve model in web based application to be used by others? ( for teaching purpose)
  • Marco_BoeckMarco_Boeck Administrator, Moderator, Employee, Member, University Professor Posts: 1,993 RM Engineering
    Hi,

    have a look at our YT channel: https://www.youtube.com/user/RapidIVideos/videos
    This (older) video might still be helpful: https://www.youtube.com/watch?v=SBCUwA6llns

    Regards,
    Marco
Sign In or Register to comment.