How do you identify a First Order Plus Dead Time (FOPDT) model from process data?

FOPDT model: Y(s)/U(s) = K*exp(-theta*s)/(tau*s + 1). Identification from step test: apply step change in manual, record response. Graphically: K = delta_output/delta_input at steady state; theta = time from step to initial response; tau = time from initial response to 63.2% of final change. Or use two-point method (t1 at 28.3%, t2 at 63.2%): tau = 1.5*(t2-t1), theta = t2 - tau. Software methods fit data using least squares. Model used for tuning calculations and control design.