alberto.otano
/
curso-series-temporales
mirror of https://github.com/albertootano/curso-series-temporales.git


  
1

	
2

	
3

	
4

	
5

	
6

	
7

	
8

	
9

	
10

	
11

	
12

	
13

	
14

	
15

	
16

	
17

	
18

	
19

	
20

	
21

	
22

	
23

	
24

	
25

	
26

	
27

	
28

	
29

	
30

	
            # Importando los datos
library(readr)
APTelectric <- read_csv("A CURSO SERIES TEMPORALES (NUEVO)/Clases Nuevas/Redes Neuronales para series/Clase 9/APTelectricity.csv",
col_types = cols(X1 = col_skip()))
library(readr)
APTelectricity <- read_csv("APTelectricity.csv")
View(APTelectricity)
# Objeto ts
myts = ts(APTelectric$watt, frequency = 288)
APTelectric=APTelectricity
# Objeto ts
myts = ts(APTelectric$watt, frequency = 288)
plot(myts)
# Ajuste del modelo
library(forecast)
fit = nnetar(myts)
# Predicciones
nnetforecast <- forecast(fit, h = 400, PI = F)
library(ggplot2)
autoplot(nnetforecast)
# Usando una variable ex?gena
fit2 = nnetar(myts, xreg = APTelectric$appliances)
# Definiendo los pron?sticos de la variable ex?gena para 10 horas
y =rep(2, times = 12*10)
nnetforecast <- forecast(fit2, xreg = y, PI = F)
autoplot(nnetforecast)
# Definiendo los pron?sticos de la variable ex?gena para 30 horas
y =rep(2, times = 12*30)
nnetforecast <- forecast(fit2, xreg = y, PI = F)
autoplot(nnetforecast)