CS代考

Semester 2 2021
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 1 / 11
Tiime Series Analysis Practice

Time Series Analysis in Practice
In theory, theory and practice are the same. In practice, they aren’t.
Theory without practice means you may not get useful results Practice without theory means you may get wrong results
Having the theory with practical experience means you should do well
Practical experience requires practise. And practise is doing time series analysis while thinking about what you are doing theory and on seeing the results seeing how practice differs from theory. And perhaps seeing you need more theory eg an alternative time series analysis methodology.
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 2 / 11

Theory is very useful
It is All Statistics
Six mathematical concepts underlying statistics – what makes it different
1 Expectation
2 Variance
3 Probability
4 Risk
5 Distributions
6 Correlation
©Fidelio Statistical Services
Tiime Series Analysis Practice
Semester 2 2021
3 / 11

What theory?
Should be able to understand – ie explain – important time series concepts, and why they are important.
For example
Stochastic process
Data generating process
Autocorrelation
Trend, seasonal and irregular. What differentiates them? Stationarity
Random Walk
Autoregression and moving average
How does a time series, and thus its analysis, differ from data?
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 4 / 11

Why Do Time Series Analysis?
The official reasons. Description Explanation
Control
Prediction The real reasons
Fun; or Profit
©Fidelio Statistical Services
Tiime Series Analysis Practice
Semester 2 2021
5 / 11

Doing Time Series Analysis For Profit
This is doing work because you have been asked to.
Your first question should be along the lines of “You want me do some time series analysis so that, …?”
What is the difference between
“I want you to do some time series analysis so that I have some numbers for my report”; and
“I want you to do some time series analysis so that we can decide whether to spend $100,000,000”
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 6 / 11

Starting the Analysis
It’s important you understand what is the purpose of the analysis and what the time series are like. Here are some starting points
What is the reason for doing the analysis? And is there more than one?
What is the definition of success?
How much time do I have?
Have I done something like this before?
Are there any expectations as to what the analysis should produce?
Who do I talk to about the data? My interim results (am I allowed interim results)?
Has someone looked at this before? Something similar?
Is someone going to be repeating (or replicating) the analysis?
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 7 / 11

Starting the Analysis
About the data
How much do I have? How many do I have?
Is anything missing? And how will I know it is missing?
How often is it measured?
Is the DGP likely to be same throughout the series?
Is the Data Measurement Process (DMP) likely to be the same throughout the series?
Is is continuous data or count data?
Are there many zeroes?
Are there likely to be oddities eg outliers, lots of zeroes?
You are unlikely to get all the answers, let alone questions, to start. And your analysis will raise questions.
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 8 / 11

The Analysis
It won’t be linear with a beginning and an end, but rather an exploration. Generally you find this order is useful even if you aren’t going to the end
Description Explanation Prediction Control
It will be an iterative approach, particularly if doing it more than once.
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 9 / 11

The Analysis
The Steps
1 Problem definition.
2 Gathering information.
3 Preliminary (exploratory) analysis.
4 Choosing and fitting models.
5 Using and evaluating a forecasting model.
6 Documenting your work
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 10 / 11

The Analysis
And now I’ll do an example.
©Fidelio Statistical Services Tiime Series Analysis Practice Semester 2 2021 11 / 11