Analytics - Time to failure - Time-series predictions

[skef]

Hello,

i want to use ThinWorx Analytics in order to implement a time-series prediction (TTF scenario). Unfortunately, this is not well documented. 

Can anyone provide me with more information, how to realize a (virtual sensor) time series prediction? I think a good point to start would be an example project. Especially a data set which I could use, to create a model by myself, in order to compare my results with the results of my my, would be useful for me.

Are there more data sets and models available than the one in the actual analytics guide?

[abarki]

Hi

What version of ThingWorx Analytics are you using?

Br.

Asia

[skef]

I'm using Version 8.2

[abarki]

- Attached is two different datasets for timeseries.

Br.

Asia

[skef]

Thank you very much.

How can I verify the results of the model?

What does input data shape fields windowsfunctionaverage, windowsfunctionderivate, windowsfunctionvolatility mean, if i want to test the generated model? 

If I want to test the model, the following error occurs:

The dataset must contain entity ID and temporal fields in order to perform TimeSeries training.

But the fields ENTITY_ID and TEMPORAL are contained in the json metadata and the corresponding data in the csv file.

As you can see, I have a lot of basic questions regarding time series prediction. Is there any documentation with a more basic example, how to implement this with thingworx analytics?

[abarki · Accepted solution]

- Did you view below links?

https://www.ptc.com/en/support/article?n=CS271176

http://support.ptc.com/help/thingworx_hc/thingworx_analytics_8/#page/thingworx_analytics_8%2Ftwxa-time-series.html

Best,

Asia