Trend in Time Series Regression

[sasgiaz]

Hello, does anyone happen to know the syntax to check signification of trend in time series regression?

I know the syntax Y=Time/influence but got confused interpreting the output. Please help, thank you in advance.

[PGStats · Accepted solution]

The estimate for the Time effect is the size of the trend (in the same units as Time, i.e. if Time is in hours,, the estimate is of change in Y per hour). The pvalue for the Time effect is the statistical significance of the trend.

[sasgiaz]

Thankyou for spending your time to answer my question..but I want to make it clear..

I attach the screenshot for output in SAS.

By reading your answer, is it true that based on the output, the p-value of 'waktu' (waktu = time) is valued more than alpha, then it can be concluded that the trend data is not significant, and the model should not be using trend?

Please correct me if I'm wrong, thank you so much.

[PGStats]

Correct. A p-value of 0.66 denotes a non significant effect, i.e. it has no predictive value. 

[sasgiaz]

Ok thank you!