before multiple linear regression Normality (Shapiro Wilk)

#1
I have meteorological data to estimate solar radiation (dependent var.) and Relative humidity, Soil temperature, Air temperature and cloudness so on.
Before regression analysis i did normality test but i have sig 0 adn this is men my data is not normally disributed. What i need to do now ?
My data is daily (4 year) and mothly (4 years) . Can i countunie to regression at this stage ? is it problem ?and i need durbin whatson and levene test for homosdasticity.
 

Dason

Ambassador to the humans
#2
No. Don't do normality testing before you run the model. The assumption of normality is on the error/residual term - not the raw data.
 
#5
I did multiple and linear regression analyzes for my thesis. I also found error results such as RMSE, MBE, and MAPE. I couldn't do the 3th question I sent above. I have regression prediction results Can I do Durbin watson, Shapiro wilk and Levene test using this prediction result only?