Unit 3- Time Series Analysis

Question 1

What is time series data?

Answer 1

Any data that is collected or recorded at regular time periods (daily, weekly, monthly)

Question 2

For time series data, what are the explanatory and response variables?

Answer 2

Explanatory is always a measure of time, plotted on the x axis.

Response is the variable of interest, plotted on the y axis.

Question 3

What are the ways we can describe time series data?

Answer 3

1. Trends: these can be upwards or downwards.
2. Structural changes: these are trends that change. e.g. first an upwards trend, then a downwards trend.
3. Cycles: this is when the data rises and falls over periods larger than a year.
4. Seasonal data: the same as cycles, except this is less than a year time period.
5. Outliers: one point is dramatically different to the rest of the data (state where the outlier is).
6. Irregular fluctuations: unable to reasonably attribute to any of the other descriptors.

Question 4

Smoothing time series data: How do you do a 3-point-moving mean and a 5-point-moving mean?

Answer 4

3: Go from the first value and draw a line from the 1st to the 3rd value. Find the mean value of these three numbers and draw it on the row down in the middle of the 3 (the 2nd one). Write the calculations below.

5: Same process except with 5 values. The mean on the row will end up being in the middle of two values because it is an uneven number.

Question 5

What are the formulas for deseasonalising data? And what do these values mean?

Answer 5

Seasonal index= Value for season/ Seasonal mean

Seasonal mean (horizontal section all means) Then divide each value in that season by the value calculated by the mean calculated.

Deseasonalised value= Actual value/ Seasonal index

*** because there are 4 quarters, if you add all seasonal index then it should equal 4

*** when you find the seasonal index, if it is 0.8, it is 80% meaning 20% under what it is supposed to be. If it is 1.07 then it is 7% over.