install.packages('readxl')
install.packages('forecast')
install.packages('fUnitRoots')
install.packages('tseries')
library(readxl)
library(forecast)
data <- read_excel(path = 'C:\\Users\\Admin\\Desktop\\新建文件夹\\价格预测--ARIMA.xlsx')
price= data ##数据的导入
pricets=ts(price,start=c(2017/1/1),frequency = 12) ##生成时间序列
pricets ##打印时间序列
plot(pricets) ##绘图
pricetsdiff=diff(pricets,differences=2) ##做一阶差分
plot.ts(pricetsdiff) ##得到一阶差分的图
library(fUnitRoots) #进行单位根检验
urdfTest(pricets)
library(tseries)
adf.test(pricetsdiff)
acf(pricetsdiff,lag.max=20)
acf(pricetsdiff,lag.max=20,plot=FALSE)
#Autocorrelations of series 'pricetsdiff2' ,by lag ##观察图形,找出自相关阶数--偏离边界值的第一个,确定p值
pacf(pricetsdiff,lag.max=20)
pacf(pricetsdiff,lag.max=20,plot=FALSE) ##偏自相关的阶数选择,得到q值
library(forecast)
pricearima=Arima(pricets,order=c(12,2,10))##参数由之前的决定
pricearima ##得到模型的系数,根据系数可以写出模型的表达式
pricearimaforecast=forecast(pricearima,h=5,level=c(90,99))
pricearimaforecast
plot(forecast(pricearimaforecast))
#或者 autoplot(pricearimaforecast)
#acf(pricearimaforecast$residuals,lag.max=20)##看图形,有没有超出边界的
#Box.test(pricearimaforecast$residuals,lag=20,type="Ljung-Box") ##看P值,是否拒绝原假设---大雨0.05
#plot.ts(pricearimaforecast$residuals)
qqnorm(pricearima$residuals);qqline(pricearima$residuals)
acf(pricearimaforecast$residuals,lag.max=20)##看图形,有没有超出边界的
Box.test(pricearimaforecast$residuals,lag=20,type="Ljung-Box") ##看P值,是否拒绝原假设---大雨0.05
plot.ts(pricearimaforecast$residuals)
时间序列预测
于 2021-10-11 19:28:13 首次发布