I'm not sure I understand your question well. In my opinion, You should regress Jones model for every industry in your sample. It means your regressed models will be in number of the industries in your sample. After Regressing, you have the required coefficients. Consistent with Jones model, some variable is from year t (in your research 2015) and others is from year t-1 (2014). by excluding Non discretionary Accruals measured on Betas, you have every sample firm's Discretionary accruals in 2015.