Time Series Flashcards
What is time series data?
Time series data is data that is colelcted on different points in time.
This is opposite of cross-sectional data, which looks at multiple variables in one period
What is the sample size of time series data?
The sample size represents the time periods on which the data has been analysed.
How can time series variables be?
- Stationary
- Non-stationary
-with a deterministic trend (trend-stationary) - stohastic trend (unit root test)
What does stationarity reffer to?
Stationarity happens when the graph is mean reverting (it goes up and down, it doesn not stay long up or down)
Why do we need to perform tests on time series data?
We need tests to decide whether a time series is
* stationary
* trend stationary (around a linear trend)
* unit root time series
Based on the time series we deal with, we need to make appropriate transformations of the model to reach correct results
What are seasonalities?
Seasonalities are periodic fluctuations
In this case we have stability in time, constant mean and variance
Stationary time series are easy to predict
What are non-stationary time series data?
It reffers to data that have means, variances and covariances that change over time.
Non stationary time series data are unpredictable and cannot be modeled
In order to provide accurate representations, non stationary data needs to be transformed into stationary data
What are the types of non-stationary data?
- Pure random walk
- Random walk with drift
- Deterministic trend
- Random walk with drift and trend
What is a unit root?
A unit root is a unit of measurement to determine how much stationarity a time series model has. Meaning that we determine the stochasticity (the randomness) of the model.
What is the Dickey Fuller test?
The dickey fuller test is a test to determine if a unit root (something that can cause issues in statistical inference) is present in the model
