Examples of Stationary Time Series
Let's remember the basics. What is stationary data? Data is stationary if its statistical properties, such as the mean, variance, and autocorrelation function, do not depend on time.
For example, white noise process is stationary. This process doesn't have any trends and other time series patterns, samples are not correlated with each other and variance equals 1 for all samples.
There is no particular type of data that will always be stationary. For example, we cannot say that purchases in an online store will necessarily be stationary (quite the opposite, mostly non-stationary). But if we start to examine accident statistics during some short period, we will most likely find that these data are stationary, and there is no apparent dependence on each other.
Most time series are non-stationary. That is why some methods allow you to transform data into stationary ones. In the next section, you will learn how to implement this.
Why does the data need to be stationary? Because only stationary data contains characteristics that must be observed when using the most popular predictive models. Moreover, even just from the point of view of logic, if the data does not obey any particular law and has different characteristics at different points of time, then such data is very difficult to predict
Everything was clear?
Course Content
Time Series Analysis
1. Time Series: Let's Start
2. Time Series Processing
3. Time Series Visualization
Time Series Analysis
Examples of Stationary Time Series
Let's remember the basics. What is stationary data? Data is stationary if its statistical properties, such as the mean, variance, and autocorrelation function, do not depend on time.
For example, white noise process is stationary. This process doesn't have any trends and other time series patterns, samples are not correlated with each other and variance equals 1 for all samples.
There is no particular type of data that will always be stationary. For example, we cannot say that purchases in an online store will necessarily be stationary (quite the opposite, mostly non-stationary). But if we start to examine accident statistics during some short period, we will most likely find that these data are stationary, and there is no apparent dependence on each other.
Most time series are non-stationary. That is why some methods allow you to transform data into stationary ones. In the next section, you will learn how to implement this.
Why does the data need to be stationary? Because only stationary data contains characteristics that must be observed when using the most popular predictive models. Moreover, even just from the point of view of logic, if the data does not obey any particular law and has different characteristics at different points of time, then such data is very difficult to predict
Everything was clear?
