I have two univariate time-series with seasonality and trend--dt1 and dt2. I believe that dt1 and dt2 are strongly correlated, both through a few statistical test (see below) and that in my field dt2 is a proxy for dt1.

Linear Regression Research

Plot (shows seasonality and trend--which was expected) Plot

Pearson correlation = .815 Scatter plot showing correlation and checked residuals for OLS enter image description here

enter image description here


Ultimately I want to use dt1 off dt2.I don't actually need to forecast dt2 or make predictions about it, since it is a leading indicator and I actually can collect that data before dt1 is available.

Since I don't need to actually predict/forecast dt1 on it's own--I don't think (and I can be completely wrong) ARIMA or lagged/autoregressive studies--I am using dt2 to forecast it.

Is a Linear Regression the best starting point (even with dt1 having obvious season/trend time components)--should I be looking into a ARIMAX or another tool?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.