Very often in our work we come across such a concept as "time series". This definition was coined a long time ago. Then, when people just began to record data about something with two meanings: a phenomenon and a time. The most classic description of a time series is the recording of temperature over a year or several years.
But the series itself is just a collection of information that does not carry anything necessary. Moreover, if you build a graph of this series, using, for example, time values for the Y-axis, and readings that were originally recorded or formatted by us in digital form for the X-axis, then we can find some sequences.
In the case of a temperature chart, the day is warmer than the night and winter is colder than summer. And the more data we can analyze in this way, highlighting some patterns, the more accurately we will be able to predict what awaits us in the future.
People in the past thought the same way, dividing the process of working with timelines into three stages: collecting data, analyzing a time series, predicting the next values.
But what can a time series be used for in auditing? For everything!
– . – . – ! ( — , )
, . , , .
– .
: .
- :
. – .
, 200? 10 , ?
, . :
;
.
, - , — .
, — . - , . , ?
:
— ,
— ,
,
— ,
— — — ,
DF — GLS,
.
— , statsmodels python. (statsmodels pandas), , . :
:
:
(H0): , , ( ).
(H1): , , . , , .
p -values .
p -values — ( ).
p –values 5%, , , , , , .
, , – .
, , , .
, .