The University of Hong Kong · FACULTY OF STATISTICS

STAT8003 Chap.1 Time Series Data and Dependence

- one subject, every graph, every model, every mark
6 Chapters3-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 1 of 8 · STAT8003

Time Series Data and Dependence

Start with the indexing contract

This chapter separates an observed time-ordered record from the stochastic process that models it.

It explains why regular spacing gives lag a consistent meaning, why lowercase observations and uppercase random variables have different roles, and how deterministic-time models differ from stochastic models that use earlier process values.

Keep data and model on different lines

A finite lowercase vector records what was observed.

The uppercase process describes the random mechanism that could have generated that vector and supports expectations, variances and joint distributions. Parameters remain fixed model quantities, while innovations represent new information arriving through time.

Naming these roles prevents a realised number from being treated as though it still had a probability distribution.

Ask what carries the signal

A deterministic-time equation uses the known index to describe systematic movement. A stochastic dependence equation uses earlier random values or shocks. They can show similar paths over a short interval but imply different moment and forecast calculations.

The final decision is directional: a causal model may use current and past information, while a future innovation cannot explain a real-time prediction.

In this chapter

What this chapter covers

  • 01

    Regularly spaced observations

  • 02

    Realisation and random process

  • 03

    Time-index notation

  • 04

    Deterministic trend model

  • 05

    Stochastic dependence model

  • 06

    Innovation information set

Worked example · free

Separate the data vector from its model

Q [6 marks]. AskSia-authored practice. The marks shown are an AskSia study allocation, not a University marking scheme. Four hourly sensor readings are 8, 11, 9 and 12. Write the observed record, the corresponding random vector and one first-order stochastic model.
The observed record has an arrow.The observed record has an arrow. Equal spacing lets a one-step lag mean the same elapsed time throughout the analysis.z1z2z3z4equal time spacingone realised record
The observed record has an arrow. Equal spacing lets a one-step lag mean the same elapsed time throughout the analysis.
  • 2Write the lowercase realised vector.
  • 2Write its uppercase random counterpart.
  • 2Give a model using the immediate past and a new innovation.
The record is (z1,z2,z3,z4)=(8,11,9,12), while the random vector is (Z1,Z2,Z3,Z4). One admissible model is Zt=theta0+phi Zt-1+at. The observations do not determine theta0, phi or the innovation law by themselves.
Sia tip — Label each symbol as observation, random variable, innovation or parameter before doing algebra.
Glossary

Key terms

Observed Record
The finite sequence of realised values collected from a process. It is written with lowercase symbols, preserves chronological order and supplies the sample evidence used to learn about an unobserved stochastic mechanism.
Stochastic Process
A family of random variables indexed by time. Its joint distributions describe how possible values and their dependence change across the index before one particular observed record is realised.
Innovation
New random information entering the process at a given time. In a causal model, current and earlier innovations may affect the present value, while future innovations remain outside the available information set.
FAQ

Time Series Data and Dependence FAQ

Why does regular spacing matter?

Regular spacing makes one lag correspond to the same elapsed time across the record. Without it, a one-row shift could mean one day in one place and a week elsewhere, so standard lag covariance and correlation calculations would mix unlike intervals.

Why are lowercase and uppercase symbols kept separate?

Lowercase z values are realised observations after collection, whereas uppercase Z variables describe the random process before realisation. Expectations and covariances belong to the uppercase model; plots and sample calculations use the observed lowercase record.

Does an ordered sequence automatically have dependent values?

No. Ordering creates the possibility and meaning of lagged dependence, but a process may still be independent. Dependence is a modelling or empirical property that must be stated, derived or estimated rather than inferred from the presence of a time index alone.

How should I check a proposed time-series model?

First identify what carries the predictable part: the time index, earlier process values or earlier innovations. Then label every symbol by role and check whether the equation uses information available at time t. Confirm that the sampling interval gives each lag a consistent duration.

Only after those checks should you derive expectations or covariances, because the same-looking coefficient has a different interpretation in a deterministic trend and an autoregressive model.

What is a useful final notation check?

Underline the time index, circle random variables and innovations, and box fixed parameters. Then read the equation aloud in time order. This exposes a future shock, an irregular lag or a realised observation that has accidentally been treated as a random model quantity.

Study strategy

Exam move

Begin every question by naming the object. Mark whether the expression uses observed values, random variables, parameters or innovations. Next check whether the time intervals are regular and whether the proposed model uses the time index or past process values as its predictable component. This sequence prevents later covariance formulas from being applied to the wrong quantity.

Build a four-column notation sheet headed observation, random variable, parameter and information. Place each new symbol into one column before manipulating it. Then practise translating a short observed vector into its random counterpart without claiming that the realised numbers determine a distribution. Use several sampling schedules and decide whether adjacent rows represent a consistent time interval.

For every irregular schedule, state exactly what would go wrong if a row shift were called one lag. Compare a deterministic trend model with an autoregressive equation using the same numerical coefficient, and explain why the coefficient describes a change per unit time in one model but persistence from the previous value in the other.

Finish with causality language: information arriving at time t can enter the current value, while a future innovation cannot support a real-time forecast. A good oral test is to explain the distinction without formulas and then recreate the formulas from that explanation. During revision, replace the course examples with a new hourly, daily or quarterly setting and check that the notation still works.

Before submitting an answer, underline the time index, circle every random quantity and box every fixed parameter. That visual audit exposes role changes that ordinary algebra can hide.

Working through Time Series Data and Dependence in STAT8003? Sia is AskSia’s AI Statistics tutor — ask any STAT8003 Time Series Data and Dependence question and get a clear, step-by-step explanation grounded in how STAT8003 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 52 of your subjects at The University of Hong Kong, plus 1,000+ more Bibles across the AskSia library.
Sia - your STAT8003 tutor, unlimited, worked the way the exam marks it
The full 3-page Bible + practice bank with worked solutions
Chrome extension - sync your LMS so Sia knows your deadlines
Bilingual EN / Chinese on every Bible and every Sia answer
$0.99 Trial
30-day money-back · cancel in one tap · how it works