Composition of time series

The main components of time series data include: trend, seasonality and residual. Seasonality is divided into additive seasonality and multiplicative seasonality. You can use the seasonal decomposition method of statsmodes package to decompose time series.

Time series is a series of data points arranged in the order of time occurrence, which contains some information. In time series data, it usually includes the characteristics of data development trend (rising, falling and maintaining) and data change law (seasonality). These characteristics often have certain regularity and predictability. Specifically, time series data has the following characteristics:

Trend: A variable shows a relatively slow and long-term upward, downward and unchanged trend with the progress of time or the change of independent variables, but the range of change may be unequal.

Seasonality: the law of peaks and valleys caused by external influences such as the alternation of natural seasons.

Randomness: individual moments present random changes, and the whole presents statistical laws.

In some application scenarios, it is necessary to forecast the time series data, such as retail, e-commerce and other industries, and it is necessary to forecast the future sales amount, passenger flow and order volume. Accurate prediction results can provide decision-making reference for enterprise leaders, help to improve the human efficiency of enterprises and bring more profits to enterprises.

The seasonality in time series data refers to the law of periodic changes contained in the data, which is often closely related to seasonal time points such as year, month, season and week. Usually seasonal factors can be divided into 1. Additional seasonality and 2. Multiplied seasonality.