ThingWorx Analytics-Manager: Question regarding the data mapping process with a Time series Data set

[DF_10027356]

Hello all,

I have some questions regarding the data mapping process in the Thingworx-Analytics-Manager for creating an Analysis event.

Here are some basic information about the predicitve model: 

- I want to predict the Time to Failure (TTF) and have followed this instruction: https://community.ptc.com/t5/IoT-Tech-Tips/Predicting-Time-To-Failure-with-ThingWorx-Analytics/m-p/579897#M1095 

- In my model I used the function "useGoalHistory" becaus this showed better results for my usecase

- the data set is a time series data set

- ThingWorx Version 9.0

My questions:

1. In the Input Data Shape my goal variable appears ("TTF"). How can this happen? I thought that I have to map the goal variable in the Result mapping? Is this maybe caused by the fact that I am using the goal history?

2. One input parameter changed the datytype. During uploading the dataset on ThingWorx Analytics for training the model, the parameter was described as "Integer". Now when I want to map this parameter to the data of a Thing the parameter suddenly has the data type "Number". My problem now: Now I can't map the parameter to the data in the Thing because the datatypes don't match. How can I solve this problem?

Thank you already in advance.

Best regards

David

[cmorfin]

Hi @DF_10027356 

Indeed you have the goal in the input datashape because you sue use goal history, this make the goal a input for the next prediction.

Regarding the change of data type, it might be good to open a case to Technical Support for further investigation.

As a test, you can try to manually change the data type in the datashape directly.
If you select your Analysis Model and select View > Datashapes, you will see the name of the datashape used.

You can then find them in Composer and change the type from Number to Integer.

Hope this helps

Regards

Christophe