Considering the analysis of time series not as an abstract statistical concept, but as a phenomenon widely used in practice, we can conclude that this topic is very relevant today to study a number of processes. It is especially in demand in human economic activity, so most of the examples in popular science literature are given precisely from the point of view of its use in this context. But this does not stop the scope of the use of studying and evaluating time series.
The very definition of a time series in many respects reminds us of the process of collecting any statistical information, and consists in clearly fixing at certain time intervals real indicators measured in a way that gives the greatest reliability. In other words, when describing any phenomenon, a graph is used where the temporal indicators of measurement are fixed on the abscissa axis, and its real physical quantities on the ordinate axis.
In fact, the methods of analyzing time series at one time formed the basis for the description of many physical laws and technical processes. Their generalization allowed the process of description to be reduced to a certain mathematical expression. But not all processes were able to fit into the framework of clear formulas. And no one has canceled the solution to two main problems. They are:
- determination of the nature of the series;
- forecasting.
So the analysis of time series received an additional incentive for its development, and a rich set of tools and methods appeared in its arsenal.
A classic example of a time series is a series proposed in 1976 by Box and Jenkins. Using the example of studying the activity of monthly international air transportation for twelve years in the period 1949-1960, they showed the presence of two components: an almost linear trend and seasonal changes. When traffic growth steadily increased, and depending on the season, areas of surge and attenuation of activity were periodically observed. This type of description is called a model with multiplicative seasonality.
In the same year, the same Boxing and Jenkins proposed a very interesting in terms of forecasting, but very laborious and complex method of Autoregressive Integrated Moving Average (ARPSS).
In the study of processes subject to influence from outside, the practical method of interrupted time series has gained distribution. It was described in the 80s of the last century. The essence of the method is to study the processes after intervention in the system from the outside. The analysis of time series was to evaluate the introduction of new leadership methods, the use of various know-how, the impact of legislative processes , etc.
Spectral analysis of time series appeared on the basis of previous methods. Among the evaluation criteria by this method, the period and frequency are clearly visible. Quite widely used in calculations are complex numbers, Fourier transforms.
The abundance of methods and methods that involves the analysis of time series confirms how fertile this soil is for further research. After all, descriptions of these processes are rather cumbersome and require some experience from the analyst. A powerful leap in the development of personal computing has led to the conclusion of this type of analysis to a new qualitative level. And the ubiquity of the Internet has made available to a wide category the results of recent research in this area.
What, if not a time series analysis, is used by a successful player in the Forex market, it is the study of enterprise development schedules that allows the manager to develop the right strategic line, and market assessment provides an extensive field of activity for marketers and managers, allowing you to adjust the price level and range of products or services sold with aim to maximize benefits.
Each analysis method deserves special attention and requires thorough study. And if at least one of them interested you, then the goal of the article has been achieved.