08. Time Series Analysis

Question 1

What is PACF

Answer 1

Partial Auto Correlation Function - shows only direct linkages i.e. day 7 is directly affected by day 0

Question 2

What is ACF

Answer 2

Auto Correlation Function - shows the indirect linkages i.e. day 1 is indirectly affected by day 0. and each day onto the next.

Question 3

What is fanovarience

Answer 3

An increasing spread of the data but around a constant mean

Question 4

What are the four parts of a decomposed time series

Answer 4

Trend - steady underlying long term trend
Seasonal - regular repeating pattern over time
Cyclic - also a periodic fluctuation but over lengthy periods ie Boom and Bust
Random - white noise

Question 5

What is the box-jenkins methodology

Answer 5

Condition the Data - check and make it stationary & see if there is seasonality
Estimate the model parameters - PACF & ACF
Run Diagnostics (repeat potentially)

Question 6

What is covariance in time series

Answer 6

Two time series mirror (or do the reverse for a negative covariance) each other

Question 7

What is autocorrelation ACF

Answer 7

Indirect relationship. This is the moving average - how one point is related to the next point (or a number of points over a wider range).

Question 8

What is partial autocorrelation PACF

Answer 8

Direct relationship. This is where something happens on days x days away from a day which manifests later. Autoregressive