Ejercicio-video

2023-01-22
#Ejecutamos los paquetes
library(readxl)
library(tseries)

## Registered S3 method overwritten by 'quantmod':

## method from
## as.zoo.data.frame zoo

library(astsa)
library(forecast)

##
## Attaching package: 'forecast'

## The following object is masked from 'package:astsa':

##
## gas

library(tidyverse)

## ── Attaching packages
## ───────────────────────────────────────
## tidyverse 1.3.2 ──

## ✔ ggplot2 3.4.0 ✔ purrr 1.0.0

## ✔ tibble 3.1.8 ✔ dplyr 1.0.10
## ✔ tidyr 1.2.1 ✔ stringr 1.5.0
## ✔ readr 2.1.3 ✔ forcats 0.5.2
## ── Conflicts ──────────────────────────────────────────
tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()

library(lubridate)

## Loading required package: timechange

##
## Attaching package: 'lubridate'
##
## The following objects are masked from 'package:base':
##
## date, intersect, setdiff, union

library(foreign)
library(quantmod)
## Loading required package: xts
## Loading required package: zoo
##
## Attaching package: 'zoo'
##
## The following objects are masked from 'package:base':
##
## as.Date, as.Date.numeric
##
##
## Attaching package: 'xts'
##
## The following objects are masked from 'package:dplyr':
##
## first, last
##
## Loading required package: TTR

#Cargamos la base de datos

Arimar= read_excel("Arimar.xlsx", sheet = "Precios")
Arimar

## # A tibble: 87 × 1
## Precio
## <dbl>
## 1 98.0
## 2 103.
## 3 101.
## 4 97.8
## 5 97.9
## 6 97.0
## 7 99.7
## 8 99.2
## 9 98.1
## 10 93.9
## # … with 77 more rows

#Programamos para que sea una serie de tiempo

Arimar.ts=ts(Arimar, start=c(2013,1), frequency=12)
Arimar.ts

## Jan Feb Mar Apr May Jun Jul

## 2013 97.98278 103.35294 100.77090 97.75538 97.90686 96.99315 99.67255
## 2014 88.31081 90.82224 91.39522 93.88863 95.08631 97.01105 93.32017
## 2015 40.45127 45.98606 46.01733 49.32555 52.03910 52.82573 45.16595
## 2016 23.10238 23.72933 28.32656 31.33582 36.72636 39.80715 37.97153
## 2017 44.50008 44.17497 41.93243 43.22118 43.85074 41.15105 43.87771
## 2018 57.45399 56.15676 57.22458 58.15973 62.89929 64.63686 66.42162
## 2019 54.05666 57.37981 59.46449 62.07703 60.33674 56.90537 57.87580
## 2020 49.79299 44.63505 28.92302
## Aug Sep Oct Nov Dec
## 2013 99.24414 98.14572 93.87329 88.48590 89.40966
## 2014 89.50745 84.12824 72.50936 63.47980 50.18138
## 2015 38.41927 36.29036 34.86077 32.50883 26.19610
## 2016 37.74090 36.82666 40.66836 38.36162 42.32596
## 2017 45.55484 48.16237 48.89114 53.34505 54.05576
## 2018 64.26567 68.36348 71.15330 59.82583 51.86867
## 2019 49.58496 55.05522 51.00463 50.68969 54.54901
## 2020

#Graficamos la serie
plot(Arimar.ts)

# Comprobamos si es estacionaria
adf.test(Arimar.ts,alternative = "stationary")

##
## Augmented Dickey-Fuller Test
##
## data: Arimar.ts
## Dickey-Fuller = -1.6185, Lag order = 4, p-value = 0.7328
## alternative hypothesis: stationary

#Dado que el valor p-value= 0.7328 es mayor que 0.05 el modelo no es

estacionario por lo tanto debemos diferenciar hasta que tenga un
comportamiento estacionario.
seriedif=diff(Arimar.ts)
seriedif
## Jan Feb Mar Apr May
## 2013 5.37015544 -2.58203596 -3.01551721 0.15147672
## 2014 -1.09885152 2.51143263 0.57298352 2.49340638 1.19767879
## 2015 -9.73010322 5.53478826 0.03127119 3.30821931 2.71355401
## 2016 -3.09371320 0.62694765 4.59723325 3.00925512 5.39054257
## 2017 2.17411883 -0.32511051 -2.24254299 1.28875229 0.62956241
## 2018 3.39823914 -1.29723298 1.06782245 0.93514799 4.73955640
## 2019 2.18799111 3.32314972 2.08468274 2.61253762 -1.74029364
## 2020 -4.75601993 -5.15794324 -15.71202716
## Jun Jul Aug Sep Oct
## 2013 -0.91370875 2.67940211 -0.42841493 -1.09842379 -4.27242329
## 2014 1.92473722 -3.69088007 -3.81271745 -5.37920515 -11.61888564
## 2015 0.78662059 -7.65977611 -6.74668310 -2.12890430 -1.42959370
## 2016 3.08078471 -1.83561950 -0.23062285 -0.91424262 3.84169565
## 2017 -2.69969198 2.72666362 1.67712769 2.60752811 0.72876576
## 2018 1.73757612 1.78475467 -2.15595370 4.09781659 2.78982197
## 2019 -3.43136360 0.97042445 -8.29083642 5.47025681 -4.05059107
## 2020
## Nov Dec
## 2013 -5.38739453 0.92376097
## 2014 -9.02955760 -13.29842500
## 2015 -2.35193927 -6.31273336
## 2016 -2.30673298 3.96433886
## 2017 4.45391253 0.71070753
## 2018 -11.32747391 -7.95716144
## 2019 -0.31493237 3.85931908
## 2020

#Graficamos la primera diferenciación

plot(seriedif)
#Comprobar si se volvio estacionario
adf.test(seriedif,alternative= "stationary")

##
## Augmented Dickey-Fuller Test
##
## data: seriedif
## Dickey-Fuller = -3.3327, Lag order = 4, p-value = 0.07162
## alternative hypothesis: stationary

#De acuerdo a la prueba de Dicky-Fuller aún no es estacionario por lo tanto

se realiza una segunda diferenciación
seriedif2=diff(Arimar.ts, differences=2)
seriedif2

## Jan Feb Mar Apr May

## 2013 -7.95219140 -0.43348124 3.16699393
## 2014 -2.02261249 3.61028415 -1.93844911 1.92042286 -1.29572759
## 2015 3.56832179 15.26489148 -5.50351707 3.27694813 -0.59466530
## 2016 3.21902016 3.72066085 3.97028560 -1.58797814 2.38128745
## 2017 -1.79022004 -2.49922933 -1.91743248 3.53129528 -0.65918988
## 2018 2.68753160 -4.69547212 2.36505543 -0.13267446 3.80440841
## 2019 10.14515255 1.13515861 -1.23846698 0.52785488 -4.35283126
## 2020 -8.61533902 -0.40192331 -10.55408391
## Jun Jul Aug Sep Oct
## 2013 -1.06518548 3.59311086 -3.10781704 -0.67000886 -3.17399950
## 2014 0.72705843 -5.61561728 -0.12183738 -1.56648770 -6.23968049
## 2015 -1.92693342 -8.44639670 0.91309301 4.61777880 0.69931060
## 2016 -2.30975786 -4.91640421 1.60499665 -0.68361976 4.75593827
## 2017 -3.32925439 5.42635559 -1.04953593 0.93040041 -1.87876235
## 2018 -3.00198029 0.04717855 -3.94070837 6.25377029 -1.30799462
## 2019 -1.69106996 4.40178805 -9.26126087 13.76109323 -9.52084788
## 2020
## Nov Dec
## 2013 -1.11497124 6.31115551
## 2014 2.58932805 -4.26886741
## 2015 -0.92234557 -3.96079409
## 2016 -6.14842864 6.27107185
## 2017 3.72514678 -3.74320500
## 2018 -14.11729588 3.37031247
## 2019 3.73565870 4.17425146
## 2020

#Graficamos la segunda diferenciación

plot(seriedif2)

#Comprobar si se volvió estacionario

adf.test(seriedif2,alternative = "stationary")

## Warning in adf.test(seriedif2, alternative = "stationary"): p-value

smaller than
## printed p-value
##
## Augmented Dickey-Fuller Test
##
## data: seriedif2
## Dickey-Fuller = -5.1673, Lag order = 4, p-value = 0.01
## alternative hypothesis: stationary

#Ahora determinaremos el número de media móviles para el modelo arima.

acf(seriedif2)

#Ahora determinaremos el número de autoregresiones para el modelo arima.

pacf(seriedif2)
#Ahora programamos la coincidencia de rezago con la frecuencia :1
acf(ts(seriedif2,frequency = 1))
pacf(ts(seriedif2,frequency=1))

#Programamos el modelo Arima

modelo1=arima(Arimar.ts, order=c(1,2,1))
modelo1

##
## Call:
## arima(x = Arimar.ts, order = c(1, 2, 1))
##
## Coefficients:
## ar1 ma1
## 0.4173 -1.000
## s.e. 0.1100 0.047
##
## sigma^2 estimated as 17.51: log likelihood = -244.07, aic = 494.14

tsdiag(modelo1)
#Verificamos el ajuste de los datos al modelo Arima
Box.test(residuals(modelo1), type = "Ljung-Box")

##
## Box-Ljung test
##
## data: residuals(modelo1)
## X-squared = 0.055051, df = 1, p-value = 0.8145

error=residuals(modelo1)
#Graficamos el error
plot(error)
#Realizamos el pronóstico con el paquete forecast:
pronostico = forecast ::forecast(modelo1, h=10, level=95)
pronostico

## Point Forecast Lo 95 Hi 95
## Apr 2020 21.856654 13.608341 30.10497
## May 2020 18.398192 4.014833 32.78155
## Jun 2020 16.445341 -3.178974 36.06966
## Jul 2020 15.120795 -9.040917 39.28251
## Aug 2020 14.058445 -14.118113 42.23500
## Sep 2020 13.105513 -18.699183 44.91021
## Oct 2020 12.198241 -22.943549 47.34003
## Nov 2020 11.310024 -26.944576 49.56462
## Dec 2020 10.429758 -30.760523 51.62004
## Jan 2021 9.552811 -34.430032 53.53565

#Graficamos
plot(pronostico)
#Programos los nombres de los ejes y el título del gráfico.
plot(pronostico, main="Pronóstico del precio del petróleo", sub= "tiempo",
ylab= "precio del petróleo")