If you have to do d-order differences on a time series to get a stationary series, then you use the ARIMA(P, D, Q) model, where D is the order of the difference. ARIMA(P, D, Q) Model is fully known as Autoregressive Integrated Moving Average Model (ARIMA). AR is Autoregressive and P is an Autoregressive term. MA is the moving average, Q is the number of moving average terms, and D is the difference times made when the time series becomes stationary.

Here are some basic ways to view help:

A. Help ()

two

1. Open R interface 2. 3. Click on “packages” in the pop-up page and then go to “…” |

3.

library(help=”MASS”)

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

data< – XTS (data,seq(as.POSIXct(“2014-01-01″),len=length(data),by=”day”)

Error in as. Vector (x, mode) :

cannot coerce type ‘closure’ to vector of type ‘any’

Solution: Just because the blogger did not replace the input data (source), it should be data< -xts(source,seq(as.POSIXct(“2014-01-01″),len=length(source),by=”day”))

acf < -acf (data_diff1,lag.max=100,plot=FALSE)

Error in na.fail. Default (as. Ts (x)) : there is a missing value

in the object

Solution: acf & lt; – acf(data_diff1,lag.max=100,na.action = na.pass,plot=FALSE)

However, in the ACF figure displayed at this time, the maximum value of the horizontal axis coordinate (hysteresis value) is not 100, and the horizontal axis coordinate grows exponentially with the value of E +00 and E +02.

After checking the specific value of ACF, it was found that the original horizontal coordinate of lag=1 was 86400, which should be changed to the unit of seconds (?). .

There is no way to solve this problem at present, just put “data< – “(data, seq (as POSIXct (” 2014-01-01″), len = length (data), by = “day”)) “this step can be omitted…

data.fit < – arima (data, order = c (7, 0), seasonal = list (order = c (1, 0), the period = 7))

Here’s a seasonal setup for ARIMA models, the setup rules are unclear.

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

The final result is the same as the original blogger

Question: Do I need to do a unit root test?(Verified to be stationary time series)

### Read More:

- VTK learning notes: visual model
- Android learning notes 03: some problems and solutions in the learning process
- OpenGL learning notes and other learning thinking
- Learning notes of OpenGL — blending
- OpenGL learning notes: Problems and Solutions
- Learning notes — opengl01
- R language notes – sample() function
- Error analysis of multiple linear regression in R language model.frame.default
- Learning notes of Python 3: debugger speedups using Python not found
- Java learning error information 3 — Java notes
- Python learning notes (5) — cross entropy error runtimeerror: 1D target tensor expected, multi target not supported
- [Mac OS] ASUS z97-k r2.0 + gtx960 + clover v2.4k r4098 install Sierra 10.12.5 problems and Solutions
- R-common errors and their possible causes — Notes
- Use of rep function in R
- ISLR reading notes (3) classification
- The function of flatten layer in deep learning
- The difference of. Pt,. PTH,. Pkl and the way to save the model
- Problems in the second day of Android learning
- The problem that the normal of the model is no longer perpendicular to the surface after unequal scaling
- Android Development notes — mediaplayer error (1, – 2147483648)