240 likes | 247 Views
Learn about probability theory, stationarity, models for time series, forecasting, spectral analysis, and more in this comprehensive time series analysis course.
E N D
Statistics 349.3(02) Analysis of Time Series
Course Information • Instructor: • W. H. Laverty • 235 McLean Hall • Tel: 966-6096 Email: laverty@math.usask.ca • Class Times • MWF 12:30-1:20pm Biol 125 • Mark break-up • Final Exam – 60% • Term Tests and Assignments – 40%
Course Outline 1. Introduction and Review of Probability Theory • Probability distributions, Expectation, Variance, correlation • Sampling distributions • Estimation Theory • Hypothesis Testing
Course Outline -continued 2. Fundamental concepts in Time Series Analysis • Stationarity • Autocovariance function, autocorrelation and partial autocorrelation function 3. Models for Stationary Time series • Autoregressive (AR), Moving average (MA), mixed Autoregressive-Moving average (ARMA) models 4. Models for Non-stationary Time series • ARIMA (Integrated Autoregressive-Moving average models)
Course Outline -continued 5. Forecasting ARIMA Processes 6. Model Identification and Estimation 7. Models for seasonal Time series 8. Spectral Analysis of Time series
Course Outline -continued 9. State-Space modeling of time series, Hidden Markov Model (HMM) • Kalman filtering 10. Multivariate (Multi-channel) time series analysis 11. Linear filtering
Some examples of Time series data
Example Measuring brain activity in an insect as an object is approaching. Time t = 0 is at the point of impact.
Simulation To aid in the understanding of time series it is useful to simulate data
Generating a random number from a distribution Let • f(x) denote the density function • F(x) denote the cumulative distribution function = P[X ≤ x] • F-1(x) denote the inverse cumulative distribution function
f(x) F(x) denote the cumulative distribution function F(x) x
f(x) u F(x) denote the cumulative distribution function F-1(u) If u is chosen at random from 0 to 1 then x = F-1(u) is chosen at random from the density f(x). In EXCEL the following function generates a random observation from a normal distribution. = NORMINV(RAND(), mean, standard deviation)
Example Random walk A random walk is a sequence of random variables {xt} satisfying: xt = xt – 1 + ut where {ut} is a sequence of independent random variables having mean 0, standard deviation s. (usually normally distributed) The excel functions • NORMINV(prob, mean, standard deviation) computes F-1(prob) for the normal distribution. • RAND() computes and random number from the Uniform distribution from 0 to 1. • NORMINV(RAND(), mean, standard deviation) computes and random number from the Normal distribution with a given mean and standard deviation.