In the remaining Labs the GSI, Apratim Guha, will present a variety of examples of two types of time series computations: time-side parametric system identification and frequency-side nonparametric system identification.
He will also pass around, for your examination, two past Statistics 248 projects that received grades of A+.
The Lab is Friday 11-12. Apratim also holds Office Hours.
The analyses are:
1. Time-side parametric: proceed via ARMA(p,q) or ARIMA(p,d,q).
Estimation of p, q, d.
If appropriate, do some predicting.
2. Frequency-side nonparametric:
estimate the power spectra, transfer function and the coherence.
Choice of window.
3. In both of cases 1. and 2. assess the fit/validate the fit by either time-side or frequency-side procedures.
There are more details re the project at:
www.stat.berkeley.edu/~brill/Stat248/structure.html
The submitted projects are expected to contain both types of analysis.