Time series editing
Composition: long-term trend, seasonal change, periodic change and irregular change.
Long-term trend (t
) the general trend of phenomena under the action of a fundamental factor in a long period of time.
Seasonal variation
) the phenomenon changes regularly and periodically with the changes of the four seasons in a year.
Periodic change (c
) the regular changes of the fluctuation pattern of the phenomenon in the cycle of several years.
Irregular change (I
) is an irregular change, including strictly random change and irregular mutation with great influence.
For example, the year in the following table is [1]
2 Element 1: editing
time
t; Gross Domestic Product (GDP)
3 Element 2: Editing
given value
age
Gross Domestic Product (GDP)
(billion yuan)
age
Gross Domestic Product (GDP)
(billion yuan)
1994
1995
1996
1997
1998
1999
48 198
60 794
7 1 177
78 973
84 402
89 677
2000
200 1
2002
2003
2004
2005
99 2 15
109 655
120 333
135 823
159 878
182
32 1
function
1.
It can reflect the development and change process of social and economic phenomena and describe the development state and result of the phenomenon.
2. We can study the development trend and speed of social and economic phenomena.
3.
We can explore the law of the development and change of phenomena and predict some social and economic phenomena.
4.
Time series can be used for comparative analysis between different regions or countries, which is also one of the important methods of statistical analysis.
Four kinds of editing
(a) absolute time series
1. time series: time series arranged by total period index.
The main characteristics of time series are:
The index values in the 1) sequence are additive.
2) The value of each index in the sequence is directly related to the length of the period it reflects.
3) The numerical values of each indicator in the series are usually obtained by continuous registration and summary.
2. Time-point series: time series arranged by time-point total indicators.
The main characteristics of time series are:
The index values in the 1) sequence are not additive.
2) There is no direct relationship between the values of each index in the series and its interval.
3) The values of each index in the sequence are usually obtained by regular one-time registration.
(2) Relative number time series
A series of time series with the same relative number index in chronological order is called relative number time series.
(3) Average time series
The average time series refers to the time series arranged by a series of similar average index.
5 Compilation Principle Editor
Ensure the comparability of index values in each period of the series.
(a) The length of the period is preferably the same.
(b) The overall scope should be consistent.
(C) the economic content of indicators should be unified
(d) The calculation method should be uniform.
(5) Calculate the comparability of price and unit of measurement.
6- Variable Feature Editing
Nonstationarity (also translated as instability, instability): that is, time series variables can not show long-term trend, and eventually tend to a constant or linear function. [2]
Time-varying fluctuation amplitude
Volatility): that is, the variance of a time series variable changes with time. These two characteristics make it very difficult to analyze time series variables effectively. [2]
Stationary time series (stationary time)
Series) refers to the time series whose statistical characteristics will not change with time. [2]
7 Analysis Method Editor
(A) the index analysis method
Through the analysis index of time series, the development and degree of the phenomenon are revealed.
(2) Analysis method of constituent factors
Through the decomposition and analysis of the factors affecting the time series, the evolution law of the phenomenon with time change is revealed.
8 analysis model editing
Time series combination model
1 addition model: Y=T+S+C+I (Y, t
Total index with the same unit of measurement) (positive or negative deviation of S, C, I C and I from the long-term trend)
2 multiplication model: y = t s c I (universal model) (y, t
Total number of indicators with the same unit of measurement) (percentage increase or decrease of S, C, I, C and I relative to the original series of indicators)
9 forecast editing
Time series prediction is mainly based on the principle of continuity. The principle of continuity means that the development of objective things conforms to the law, and the development of things is carried out according to its own internal laws. Under certain conditions, the basic development trend of things will continue in the future as long as the conditions under which the law works do not change qualitatively.
Time series forecasting is to use statistical techniques and methods to find out the evolution model from the time series of forecasting indicators, establish a mathematical model, and make a quantitative estimate of the future development trend of forecasting indicators. [3]