Hi Guys,

I try to investigate wether Google Search data can improve a forecasting model for the German housing market regarding prices and transaction volume.

In the basemodel I would just use lagged values of the housing prices and sales as independent variables and then add the google search data to see if the errors get smaller. In a next step I include economic variables to see if the google search data is still significant or if it just proxies these economic variables.

To do that I cant use a simple regression model due to the autocorrelation with the lagged values right? So I would use an autoregressive model to conduct the forecasts and since I have to calculate these forecasts on a regional level I would use vectors (for the dependent variable all the different regions for example) in order to perform the forecast on a regional level.

Since I am not very experienced in statistics I wanted to ask you guys if that sounds reasonable or if I need to use an other technique to perform these forecasts?

Thanks in advance,