Series de Tiempo

Actividad en clase Series de Tiempo

Concepto

Una serie de tiempo es una colección de observaciones sobre un determinado fenómeno, efectuadas en momentos sucesivos, usualmente equiespaciados.

Algunos ejemplos de series de tiempo son:
1. Precios de acciones
2. Niveles de inventario
3. Rotación de personal
4. Ventas
5. PIB (GDP)

Más información:
Libro R for Data Science (2e)

Instalar paquetes y llamar librerías

if(!require('forecast')){
    install.packages("forecast")
}
library(forecast)
library(dplyr)
library(magrittr)
library(tidyr)

Crear la serie de tiempo

Ejemplo:
Los siguientes datos de producción trimestral inician en el primer trimestre de 2020.
Se busca pronosticar la producción de los siguientes 5 trimestres.

produccion <- c(50,53,55,57,55,60)
ts <- ts(data=produccion,start=c(2020,1),frequency = 4)
ts

##      Qtr1 Qtr2 Qtr3 Qtr4
## 2020   50   53   55   57
## 2021   55   60

Crear modelo ARIMA

ARIMA significa AutoRegressive Integrated Moving Average o Modelo Autorregresivo Integrado de Promedio Móvil.

arima <- auto.arima(ts, D=1)
arima

## Series: ts 
## ARIMA(0,0,0)(0,1,0)[4] with drift 
## 
## Coefficients:
##        drift
##       1.5000
## s.e.  0.1768
## 
## sigma^2 = 2.01:  log likelihood = -2.84
## AIC=9.68   AICc=-2.32   BIC=7.06

summary(arima)

## Series: ts 
## ARIMA(0,0,0)(0,1,0)[4] with drift 
## 
## Coefficients:
##        drift
##       1.5000
## s.e.  0.1768
## 
## sigma^2 = 2.01:  log likelihood = -2.84
## AIC=9.68   AICc=-2.32   BIC=7.06
## 
## Training set error measures:
##                      ME      RMSE       MAE        MPE      MAPE       MASE
## Training set 0.03333332 0.5787923 0.3666667 0.03685269 0.6429133 0.06111111
##                    ACF1
## Training set -0.5073047

Generar el pronóstico

pronostico <- forecast(arima,level = c(95),h=5)
pronostico

##         Point Forecast    Lo 95    Hi 95
## 2021 Q3             61 58.22127 63.77873
## 2021 Q4             63 60.22127 65.77873
## 2022 Q1             61 58.22127 63.77873
## 2022 Q2             66 63.22127 68.77873
## 2022 Q3             67 63.07028 70.92972

plot(pronostico)

Actividad 2 - Hershey’s

Crear la serie de tiempo

lechita <- read.csv('Ventas_Históricas_Lechitas.csv')
lechita$Ventas <- as.numeric(lechita$Ventas)
ts1 <- ts(data=lechita$Ventas,start=c(2017,1),frequency = 12)
ts1

##           Jan      Feb      Mar      Apr      May      Jun      Jul      Aug
## 2017 25520.51 23740.11 26253.58 25868.43 27072.87 27150.50 27067.10 28145.25
## 2018 28463.69 26996.11 29768.20 29292.51 29950.68 30099.17 30851.26 32271.76
## 2019 32496.44 31287.28 33376.02 32949.77 34004.11 33757.89 32927.30 34324.12
##           Sep      Oct      Nov      Dec
## 2017 27546.29 28400.37 27441.98 27852.47
## 2018 31940.74 32995.93 32197.12 31984.82
## 2019 35151.28 36133.07 34799.91 34846.17

Crear modelo ARIMA

arima1 <- auto.arima(ts1)
arima1

## Series: ts1 
## ARIMA(1,0,0)(1,1,0)[12] with drift 
## 
## Coefficients:
##          ar1     sar1     drift
##       0.6383  -0.5517  288.8980
## s.e.  0.1551   0.2047   14.5026
## 
## sigma^2 = 202700:  log likelihood = -181.5
## AIC=371   AICc=373.11   BIC=375.72

summary(arima1)

## Series: ts1 
## ARIMA(1,0,0)(1,1,0)[12] with drift 
## 
## Coefficients:
##          ar1     sar1     drift
##       0.6383  -0.5517  288.8980
## s.e.  0.1551   0.2047   14.5026
## 
## sigma^2 = 202700:  log likelihood = -181.5
## AIC=371   AICc=373.11   BIC=375.72
## 
## Training set error measures:
##                    ME    RMSE      MAE        MPE      MAPE       MASE
## Training set 25.22163 343.863 227.1699 0.08059942 0.7069541 0.06491041
##                   ACF1
## Training set 0.2081043

Generar el pronóstico

pronostico1 <- forecast(arima1,level = c(95),h=12)
pronostico1

##          Point Forecast    Lo 95    Hi 95
## Jan 2020       35498.90 34616.48 36381.32
## Feb 2020       34202.17 33155.29 35249.05
## Mar 2020       36703.01 35596.10 37809.92
## Apr 2020       36271.90 35141.44 37402.36
## May 2020       37121.98 35982.07 38261.90
## Jun 2020       37102.65 35958.91 38246.40
## Jul 2020       37151.04 36005.74 38296.35
## Aug 2020       38564.65 37418.71 39710.59
## Sep 2020       38755.23 37609.03 39901.42
## Oct 2020       39779.03 38632.73 40925.33
## Nov 2020       38741.63 37595.29 39887.97
## Dec 2020       38645.86 37499.50 39792.22

plot(pronostico1)

Finanzas Corporativas

Equipo 9

José Gabriel Usiña Mogro A00831435
Lorena Villarreal Vega A01720802
Ximena Solís Islas A00831371

Instalar paquetes y llamar librerías

if(!require('finreportr')){
    install.packages("finreportr")
}
library(finreportr)

Información disponible

Con la función finreportr podemos obtener la siguiente información:
- CompanyInfo() = Brinda información general como nombre, ubicación, ZIP, etc.
- AnnualReports() = Brinda el nombre, fecha y número de acceso
- GetIncome() = Brinda el estado de resultados
- GetBalanceSheet() = Brinda el Balance General
- GetCashFlow() = Brinda el Flujo de Efectivo

options(HTTPUserAgent = "a a@gmail.com")
CompanyInfo('JPM')

##               company        CIK  SIC state state.inc FY.end     street.address
## 1 JPMORGAN CHASE & CO 0000019617 6021    NY        DE   1231 383 MADISON AVENUE
##          city.state
## 1 NEW YORK NY 10017

AnnualReports('JPM')

##    filing.name filing.date         accession.no
## 1         10-K  2024-02-16 0000019617-24-000225
## 2         10-K  2023-02-21 0000019617-23-000231
## 3         10-K  2022-02-22 0000019617-22-000272
## 4         10-K  2021-02-23 0000019617-21-000236
## 5         10-K  2020-02-25 0000019617-20-000257
## 6         10-K  2019-02-26 0000019617-19-000054
## 7         10-K  2018-02-27 0000019617-18-000057
## 8         10-K  2017-02-28 0000019617-17-000314
## 9         10-K  2016-02-23 0000019617-16-000902
## 10        10-K  2015-02-24 0000019617-15-000272
## 11        10-K  2014-02-20 0000019617-14-000289
## 12        10-K  2013-02-28 0000019617-13-000221
## 13        10-K  2012-02-29 0000019617-12-000163
## 14        10-K  2011-02-28 0000950123-11-019773
## 15        10-K  2010-02-24 0000950123-10-016029
## 16        10-K  2009-03-02 0000950123-09-003840
## 17        10-K  2008-02-29 0001193125-08-043536
## 18        10-K  2007-03-01 0000950123-07-003015
## 19      10-K/A  2006-08-03 0000950123-06-009854
## 20      10-K/A  2006-06-22 0000950123-06-008545
## 21        10-K  2006-03-09 0000950123-06-002875
## 22      10-K/A  2005-06-28 9999999997-05-030472
## 23        10-K  2005-03-02 0000950123-05-002539
## 24      10-K/A  2004-06-28 9999999997-04-028305
## 25        10-K  2004-02-18 0000950123-04-002022
## 26      10-K/A  2003-06-30 9999999997-03-028080
## 27        10-K  2003-03-19 0000950123-03-002985
## 28      10-K/A  2002-06-28 9999999997-02-037872
## 29     10-K405  2002-03-22 0000950123-02-002823
## 30     10-K405  2001-03-22 0000950123-01-002499
## 31      10-K/A  2000-06-28 0000019617-00-000125
## 32        10-K  2000-03-13 0000950123-00-002204
## 33      10-K/A  1999-06-29 0000950123-99-006055
## 34     10-K405  1999-03-11 0000950123-99-002057
## 35     10-K405  1998-03-30 0000950123-98-003043
## 36        10-K  1997-03-25 0000950123-97-002412
## 37     10-K405  1996-03-20 0000950123-96-001197
## 38      10-K/A  1995-06-26 0000019617-95-000080
## 39     10-K405  1995-03-27 0000950123-95-000706
## 40      10-K/A  1994-05-09 0000019617-94-000048
## 41        10-K  1994-03-25 0000950123-94-000607

google_income <- GetIncome('GOOG',2016)
amazon_balance <- GetBalanceSheet('AMZN',2015)
apple_cashflow <- GetCashFlow('AAPL',2014)

Conseguir los datos

options(HTTPUserAgent = "a a@gmail.com")
jpm_income_2015<-GetIncome("JPM",2015)
jpm_balance_2015 <- GetBalanceSheet("JPM",2015)
jpm_income_2018<-GetIncome("JPM",2018)
jpm_balance_2018 <- GetBalanceSheet("JPM",2018)
jpm_balance_2013<-GetBalanceSheet("JPM",2013)
jpm_balance_2016<-GetBalanceSheet("JPM",2016)

Manipulación de Datos

conjunto<-rbind(jpm_balance_2015,jpm_balance_2018,jpm_income_2015,jpm_income_2018,jpm_balance_2013,jpm_balance_2016)
conjunto<-conjunto %>%
  filter(Metric %in% c("Assets","Net Income (Loss) Attributable to Parent","Liabilities","Deposits"))

conjunto$endDate <- as.Date(conjunto$endDate)

conjunto_filtered <- conjunto %>%
  filter(endDate >= as.Date("2012-01-01") & endDate <= as.Date("2017-12-31")) %>%
  group_by(Metric, year = as.numeric(format(endDate, "%Y"))) %>%
  mutate(row_number = row_number()) %>%
  filter(row_number == 1) %>%
  select(-row_number) %>%
  ungroup()

conjunto_filtered<-conjunto_filtered%>%select(1,3,6)

df_pivot <- conjunto_filtered %>%
  spread(key = Metric, value = Amount)

# Opcional: Renombramos las columnas para que tengan nombres más descriptivos
names(df_pivot)[1] <- "Year"

df_pivot$Assets<-as.numeric(df_pivot$Assets)
df_pivot$Deposits<-as.numeric(df_pivot$Deposits)
df_pivot$Liabilities<-as.numeric(df_pivot$Liabilities)
df_pivot$`Net Income (Loss) Attributable to Parent`<-as.numeric(df_pivot$`Net Income (Loss) Attributable to Parent`)

Crear serie de tiempo

ts2<-ts(df_pivot$Deposits,start =c(2012,1),frequency =1)
ts2

## Time Series:
## Start = 2012 
## End = 2017 
## Frequency = 1 
## [1] 1.193593e+12 1.287765e+12 1.363427e+12 1.279715e+12 1.375179e+12
## [6] 1.443982e+12

Crear Modelo ARIMA

ARIMA significa AutoRegressive Integrated Moving Average o Modelo Autorregresivo Integrado de Promedio Móvil

arima3<-auto.arima(ts2,d=1)
summary(arima3)

## Series: ts2 
## ARIMA(0,1,0) 
## 
## sigma^2 = 7.09e+21:  log likelihood = -132.88
## AIC=267.75   AICc=269.09   BIC=267.36
## 
## Training set error measures:
##                       ME        RMSE         MAE      MPE     MAPE     MASE
## Training set 41930432067 76865093443 69834432067 3.021252 5.201737 0.835714
##                    ACF1
## Training set -0.3972446

Generar el pronóstico

pronostico3 <- forecast(arima3,level=c(95), h=5)
pronostico3

##      Point Forecast        Lo 95        Hi 95
## 2018   1.443982e+12 1.278950e+12 1.609014e+12
## 2019   1.443982e+12 1.210592e+12 1.677372e+12
## 2020   1.443982e+12 1.158138e+12 1.729826e+12
## 2021   1.443982e+12 1.113918e+12 1.774046e+12
## 2022   1.443982e+12 1.074959e+12 1.813005e+12

plot(pronostico3)