homi <- read.csv ("C:/Users/jony2/OneDrive/Documentos/maestria/MATERIAS/regresion/TSLA.csv")
homi.ts <- ts(homi[,2], frequency=12, start=c(2021,2), end=c(2023,2))
#GRAFICA
autoplot(homi.ts) +
geom_line(color='black', lwd=1) +
labs(title='PRECIO ACCION TESLA',
x='FEBRERO 2021-FEBRERO 2023',
subtitle='',
caption='Fuente: Elaborado a partir de datos obtenidos de INVESTING.')
#DESCOMPOCISION
autoplot(homi.ts.des <- decompose(homi.ts)) +
geom_line(color='black', lwd=1) +
labs(title='PRECIO ACCION TESLA',
subtitle='Descomposicion clasica',
caption = 'Fuente: Elaborado a partir de datos obtenidos de INVESTING.',
x='', y='')
## Warning: Removed 6 row(s) containing missing values (geom_path).
#FAC y FACP, PRUEBA DE ESTACIONARIEDAD EN TENDENCIA Y GRAFICOS DE AUTOCORRELACION
ggAcf(homi.ts) +
geom_line() +
labs(title='PRECIO ACCION TESLA',
subtitle='Funcion de autocorrelacion',
caption = 'Fuente: Elaborado a partir de datos obtenidos de INVESTING.',
x='Retardo (k)', y='FAC')
ggPacf(homi.ts) +
geom_line() +
labs(title='PRECIO ACCION TESLA',
subtitle='Funcion de autocorrelacion Parcial',
caption = 'Fuente: Elaborado a partir de datos obtenidos de INVESTING',
x='Retardo (k)', y='FACP')
# ESTADISTICOS DE TENDENCIA Y PROCESOS DE DIFERENCIAS
homi.ts.dif <- diff(homi.ts)
fac.homi.dif1 <- ggAcf(homi.ts.dif)
facp.homi.dif1 <- ggPacf(homi.ts.dif)
ggarrange(fac.homi.dif1, facp.homi.dif1, ncol=2)
adf.test(homi.ts.dif)
##
## Augmented Dickey-Fuller Test
##
## data: homi.ts.dif
## Dickey-Fuller = -4.1883, Lag order = 2, p-value = 0.01669
## alternative hypothesis: stationary
kpss.test(homi.ts.dif)
## Warning in kpss.test(homi.ts.dif): p-value greater than printed p-value
##
## KPSS Test for Level Stationarity
##
## data: homi.ts.dif
## KPSS Level = 0.14718, Truncation lag parameter = 2, p-value = 0.1
ts_plot(homi.ts.dif, line.mode = 'lines+markers', color='black',
Xtitle = 'PERIODO ANUAL', Ytitle = 'PRECIO ACCION TESLA',
title = 'FEBRERO 2021-FEBRERO 2023')
# ESTACIONALIDAD
ggseasonplot(homi.ts)
ggsubseriesplot(homi.ts) +
ylab("PRECIO ACCION TESLA") +
ggtitle("FEBRERO 2021-FEBRERO 2023")
# SUPOCISION DE EC ARIMA
auto.arima(homi.ts)
## Series: homi.ts
## ARIMA(1,0,0) with non-zero mean
##
## Coefficients:
## ar1 mean
## 0.7475 248.6672
## s.e. 0.1249 29.2856
##
## sigma^2 = 1789: log likelihood = -128.45
## AIC=262.91 AICc=264.05 BIC=266.57
ec1 <- Arima(homi.ts, order=c(1,0,0))
summary(ec1)
## Series: homi.ts
## ARIMA(1,0,0) with non-zero mean
##
## Coefficients:
## ar1 mean
## 0.7475 248.6672
## s.e. 0.1249 29.2856
##
## sigma^2 = 1789: log likelihood = -128.45
## AIC=262.91 AICc=264.05 BIC=266.57
##
## Training set error measures:
## ME RMSE MAE MPE MAPE MASE ACF1
## Training set 1.017906 40.5644 28.47421 -2.488279 11.80674 0.3087124 0.1188868
checkresiduals(ec1)
##
## Ljung-Box test
##
## data: Residuals from ARIMA(1,0,0) with non-zero mean
## Q* = 2.5805, df = 4, p-value = 0.6303
##
## Model df: 1. Total lags used: 5
#VALIDACION EC ARIMA ELEGIDA
fit3 <- Arima(homi.ts, order=c(1,0,0))
checkresiduals(fit3)
##
## Ljung-Box test
##
## data: Residuals from ARIMA(1,0,0) with non-zero mean
## Q* = 2.5805, df = 4, p-value = 0.6303
##
## Model df: 1. Total lags used: 5
summary(fit3)
## Series: homi.ts
## ARIMA(1,0,0) with non-zero mean
##
## Coefficients:
## ar1 mean
## 0.7475 248.6672
## s.e. 0.1249 29.2856
##
## sigma^2 = 1789: log likelihood = -128.45
## AIC=262.91 AICc=264.05 BIC=266.57
##
## Training set error measures:
## ME RMSE MAE MPE MAPE MASE ACF1
## Training set 1.017906 40.5644 28.47421 -2.488279 11.80674 0.3087124 0.1188868
homi.ts.res <- fit3$residuals
autoplot(homi.ts.res) + xlab("AÑO") + ylab("") +
ggtitle("RESIDUALES")
gghistogram(homi.ts.res) + ggtitle("Histograma de los residuales")
## Warning: Using `bins = 30` by default. Pick better value with the argument
## `bins`.
## Don't know how to automatically pick scale for object of type ts. Defaulting to continuous.
ggAcf(homi.ts.res) + ggtitle("ACF of residuals")
