Is there a way to use the csv file created using the thingworx csv extension into thingworx analytics?

[jkrishna-2]

I tried to use the csv extension to export the data from value stream in thingworx platform to csv file in the repository.If someone could help me out in the steps to access this file from thingworx analytics?

[CRArko]

Hello, Janani Krishna.

I have asked one of our ThingWorx Analytics specialist to help us out with this. We should be hearing from them shortly.

Thanks,

-- Craig A.

[wposner-2]

You can't simply export a CSV file an expect to be able to import directly in to Analytics.  You first need to create your dataset config file.  This file is a JSON file which describes the data you are uploading.  The config file is required to upload any dataset in to Analytics, specifically if you're trying to upload using the Analytics Builder UI.  If you write your own services against the APIs available to you on the DefaultML thing which gets installed with the analytics extension, then you can create the JSON in javascript, format your data in an infotable, and then use the APIs to submit your data to Analytics.

I'd suggest taking a look at the sample analytics data that comes with the extension, along with the API guide, to see what  JSON looks like and what the valid data types are.  You should then be able to easily construct the services you need. 

[jkrishna-2]

I have the required csv and json data and I am able to go forward with the predictive scoring, when i get into prescriptive scoring I am trying to find the features involved in making the heuristic score change there is no response when i make the API call. Is there a way to check like logs or even the job getting created in UI like the predictive score.

Also Is it possible to get a mashup running for the predictive data score?

[wposner-2]

You should get some sort of response back from the engine.  You're doing this via Postman?  The log files are kinda hard to find in the analytics VM and I don't remember exactly where they are. 

As for a mashup for predictive scoring, yes, you can.  You just need to call the GetPredictiveResult service and then use the values to populate whatever widgets you have on your MU.  You can display your score results in a data grid or you can process your prediction value to return some sort of phrase based on the value taking in to account +/- the error.  How you display your predictive results is dictated by your use case.

[jkrishna-2]

I am using the thingworx analytics REST api calls as displayed below.

It is just in prescriptive scoring that I am not getting any response back. I just get status 202 accepted. How is it done using postman?, I tried giving the same POST method and also the parameters like the neuron application id.

[wposner-2]

You would submit to:

{{Host}}/1.0/datasets/{{Dataset_Name}}/prescriptive_scores

with the following body:

{     "description": null,

       "resultId": {{predictiveModelID}}

,      "filter": "",   

       "prescriptiveFeatures": [ ], 

       "maximize": true

}

Hopefully this works for you because I'm getting ready to write some code for prescriptive scoring.  I wouldn't be surprised of an issue in the extension service for submitting a prescriptive scoring job, but if there is an issue with the REST API, that's a problem.  What build of Analytics are you using?