none
Overflow Error while Opening Mining Model Viewer OR executing DMX query

    Question

  • Hi Everyone,

    I developed a time series model that processes succesfully. The problem I have is that when I try to view the mining model viewer that displays the times series on a chart, I get the error below:

    An error occurred while a prediction query was being executed:

    'Error (Data mining): An overflow was encountered while converting a predicted value to the '' column, at line 1, column 41. A higher precision data type for the column is recommended.'.

    I'm getting the same error while executing a DMX "SELECT FLATTENED" query on the same model.

    I didn't find any useful information browsing on the web about this error, does anyone have an idea?

    Thank you


    • Edited by Luca Seu Tuesday, October 22, 2013 10:28 AM
    Tuesday, October 22, 2013 10:16 AM

Answers

All replies

  • I guess that you model is bad. Which time series algorithm are you using? ARTXP can be very unstable which can cause this error. You can try creating a model using ARIMA.

    Tatyana Yakushev [PredixionSoftware.com]

    Download Predixion Insight 3.0 - World class predictive platform for big data

    Wednesday, October 23, 2013 4:20 PM
    Answerer
  • Thank you for your guess!

    I'm now using mixed mode, so maybe, that this is the problem. As I'll be back to the office I'll try to swhitch to ARIMA only to see if this fixes the problem.

    For now, thank you!


    Wednesday, October 23, 2013 6:09 PM
  • You were right,

    using only ARIMA the problem disappears.

    The following question anyhow remains opened: Why doesn't the algorithm already have some sort of control in order to avoid this instability? And why is the error message so obscure?

    Thank you!

    Thursday, October 24, 2013 10:04 AM