Forecasting with ARIMA models This version: 14 January 2018

Size: px
Start display at page:

Download "Forecasting with ARIMA models This version: 14 January 2018"

Transcription

1 Forecasting with ARIMA models This version: 14 January 2018 Notes for Intermediate Econometrics / Time Series Analysis and Forecasting Anthony Tay Elsewhere we showed that the optimal forecast for a mean square forecast error minimizer is the conditional mean of the target variable, conditioned on the information set being used to generate the forecast. We explore in this note forecasts based on the past history of the series itself, generated from ARIMA and trend-stationary models, assuming mean square forecast error loss functions. We assume initially that we know the DGP, and the parameters of the model. This is an unrealistic assumption, of course, but simplifies arguments so that we can focus on the properties of forecasts and forecast errors in the best case scenario. AR(1) Suppose Y t = β 0 + β 1 Y t 1 + ɛ t, ɛ t iid (0, σ 2 ), β 1 < 1, Suppose at time T, we wish to make a forecast of the realization of Y T +h. We refer to h as the forecast horizon. For h = 1, the appropriate forecast rule is obvious: Y T +1 T = E[Y T +1 Y T,...] = β 0 + β 1 Y T with forecast error ˆɛ T +1 T = ɛ T +1. For h = 2, we have Y T +2 = β 0 + β 1 Y T +1 + ɛ T +2 = β 0 (1 + β 1 ) + β1y 2 T + ɛ T +2 + β 1 ɛ T +1 Therefore Y T +2 T = E[Y T +2 Y T,...] = β 0 (1 + β 1 ) + β1y 2 T with forecast error ˆɛ T +2 T = ɛ T +2 + β 1 ɛ T +1. Notice that you will have obtained the same result by replacing Y T +1 with Y T +1 T : Y T +2 T = E[Y T +2 Y T,...] = β 0 + β 1 E[Y T +1 Y T,...] + E[ɛ T +2 Y T,...] = β 0 + β 1 Y T +1 T = β 0 (1 + β 1 ) + β1y 2 T We can use replace future unknowns by their forecasts. 1

2 Intermediate Econometrics / Time Series Analysis and Forecasting 2 We can generalize these arguments to h-step ahead forecasts. It should be clear that Y T +h T = β 0 (1 + β 1 + β β h 1 1 ) + β h 1 Y T with forecast error ˆɛ T +h T = ɛ T +h + β 1 ɛ T +h β h 1 ɛ T. Notice that as h, we have Y T +h T β 0 i=0 β i 1 = β 0 1 β 1 as h which is the conditional mean of the AR(1) process. The intuition is that information available at period T becomes less and less informative. In the limit, the best forecast we can make is one with no information from predictors, i.e., the unconditional mean. It is also straightforward to show that if β 1 is positive (we continue to assume it is less than one) then Y T +h T converges monotonically to the unconditional mean as h : if Y T > β 0 /(1 β 1 ), then Y T +h T will be decreasing in h, converging toward β 0 /(1 β 1 ). If Y T < β 0 /(1 β 1 ), then Y T +h T will be rising in h, converging toward β 0 /(1 β 1 ). As the forecast converges to the unconditional mean, it should not be surprising that the the forecast error variance converges to the unconditional variance var[ˆɛ T +h T ] σ 2 ( β1 2i ) = σ2 1 β1 2 i=0 as h. The series Y in the file timeseries_quarterly.csv is quarterly frequency from 1980Q1 to 2011Q4. library(tidyverse) library(xts) library(forecast) library(ggfortify) library(gridextra) library(grid) # Theme settings for figures: ts_thm <- theme(text = element_text(size=14), axis.title = element_text(size=11), axis.line = element_line(linetype = 'solid'), panel.background = element_blank()) # We use data from "timeseries_quarter.csv" and timeseries_monthly.csv" df_qtr <- read_csv("timeseries_quarterly.csv") glimpse(df_qtr)

3 Intermediate Econometrics / Time Series Analysis and Forecasting 3 Observations: 128 Variables: 3 $ Period <chr> "1980Q1", "1980Q2", "1980Q3", "1980Q4", "1981Q1", " $ Y <dbl> , , , , , $ Y1 <dbl> , , , , , Y.xts <- xts(df_qtr$y, order.by = as.yearqtr(df_qtr$period)) Y.ts <- as.ts(y.xts, start=start(y.xts)) autoplot(y.ts) + ts_thm + theme(aspect.ratio = 1/2) We will suppose we are are the end of 2004, and generate a long-range forecast of up to 2011Q4, without information about the realizations from 2005 onwards. fit1 <- Arima(window(Y.ts, start=c(1980,1), end=c(2004,4)), order=c(1,0,0)) summary(fit1) Series: window(y.ts, start = c(1980, 1), end = c(2004, 4)) ARIMA(1,0,0) with non-zero mean Coefficients: ar1 mean s.e sigma^2 estimated as : log likelihood= AIC= AICc= BIC= Training set error measures: ME RMSE MAE MPE MAPE MASE Training set ACF1 Training set

4 Intermediate Econometrics / Time Series Analysis and Forecasting 4 sse <- sum(fit1$residuals^2) sst <- sum((fit1$x - mean(fit1$x))^2) Rsqr <- 1 - sse/sst print(paste0("r-sqr is ", as.character(round(rsqr,2)))) [1] "R-sqr is 0.81" fcst <- forecast(fit1, h=7*4) # forecast 1 to 28 steps (7 years x 4 quarters) ahead autoplot(ts.union(y.ts, fcst$mean, fcst$lower, fcst$upper)) + scale_color_manual(values=rep("black", 6)) + ylab("") + xlab("") + aes(linetype=series) + scale_linetype_manual(values=c("solid", "dashed", rep("dotted",4))) + ts_thm + theme(legend.position="none") We do not show or analyze the fit of the model, though we present the model summary, and the (in-sample) R 2 which is The forecasts (dashed line) are shown against the actual realizations (solid) and the 80% intervals (inner dotted) and 95% intervals (outer dotted). The monotonic convergence of the forecast toward the unconditional mean is clear, as is the convergence of the forecast variance toward the unconditional variance, as indicated by the prediction intervals. The following code produces a series of one-step ahead forecasts # 1980Q1 to 2004Q4 is observation 1:100 # 2005Q1 to 2011Q4 is observation 101 to 128 (28 observations) fcst1step<-ts(matrix(rep(na,28*3),ncol=3), start=c(2005,1), frequency=4) # to store forecasts colnames(fcst1step) <- c("mean", "lower", "upper") for (i in 1:28){ fit1 <- Arima(Y.ts[1:(100+i-1)], order=c(1,0,0))

5 Intermediate Econometrics / Time Series Analysis and Forecasting 5 temp <- forecast(fit1, h=1) fcst1step[i,]<-cbind(temp$mean, temp$lower[,"95%"], temp$upper[,"95%"]) } ts.plot <- ts.union(actual=window(y.ts,start=c(2002,1)),fcst1step) autoplot(ts.plot) + scale_color_manual(values=rep("black", 4)) + ylab("") + xlab("") + aes(linetype=series) + scale_linetype_manual(values=c("solid", "dashed", rep("dotted",2))) + ts_thm + theme(legend.position="none") f1err <- ts.union(actual=window(y.ts,start=c(2005,1)),fcst=fcst1step[,1]) sse <- sum((f1err[,"actual"]-f1err[,"fcst"])^2) sst <- sum((f1err[,"actual"]-mean(f1err[,"actual"]))^2) OOSR2 <- 1-sse/sst print(paste0("out-of-sample RMSE is ",as.character(round(sqrt(sse/28),2)))) [1] "Out-of-sample RMSE is 0.54" print(paste0("out-of-sample R-sqr is ",as.character(round(oosr2,2)))) [1] "Out-of-sample R-sqr is 0.65" It is not surprising that the forecast performance looks a lot better than in the previous case, as we are only forecasting 1-step ahead, updating our information set each period (we also update the parameter estimates). As a quick evaluation of the 1-step-ahead forecasts, we can compute the

6 Intermediate Econometrics / Time Series Analysis and Forecasting 6 out-of-sample Root Mean Square Forecast Error (RMSE) by RMSE = 1 N N ˆɛ 2 T +i T +i 1 i=1 which turns out to be 0.54, highly than the in-sample RMSE of For stationary time series, it is informative to report the out-of-sample Forecast R 2 defined by Forecast-R 2 = 1 Ni=1 ˆɛ 2 T +i T +i 1 Ni (Y T +i Y ) 2 which you can compare with the in-sample R 2. Note that despite knowing the true model, and updating the fit each period, the out-of-sample Forecast R 2 is 0.65, which is lower than the in-sample R 2 of The model explains 81% of the variation in Y, but is able out-of-sample to forecast only 65% of the total variation in Y. Stochastic Trend vs Deterministic Trend For the stochastic trend model, the equations are Y T +1 = β 0 + Y }{{ T + ɛ } T +1 }{{} Y T +1 T ˆɛ T +1 T Y T +2 = β 0 + Y T +1 + ɛ T +2 = 2β 0 + Y }{{ T + ɛ } T +1 + ɛ T +2 }{{} Y T +2 T ˆɛ T +2 T Y T +3 = β 0 + Y T +2 + ɛ T +3 = 3β 0 + Y }{{ T + ɛ } T +1 + ɛ T +2 + ɛ T +3 }{{} Y T +3 T ˆɛ T +3 T The characteristics of stochastic trend forecasts are easily surmised from these equations: from Y T, the random walk with drift will forecast a linear trend (with drift term β 0 as the slope), starting from Y T, and with forecast errors that have variances increasing without bound. The forecasts from a deterministic linear trend are different from those of the stochastic trend in a nunber of ways: Y T +1 = β 0 + β 1 (T + 1) + ɛ T +1 }{{}}{{} Y T +1 T ˆɛ T +1 T Y T +2 = β 0 + β 1 (T + 2) + ɛ T +2 = β 0 + β 1 (T + 2) + ɛ T +2 }{{}}{{} Y T +2 T ˆɛ T +2 T

7 Intermediate Econometrics / Time Series Analysis and Forecasting 7 Y T +3 = β 0 + β 1 (T + 3) + ɛ T +3 = β 0 + β 1 (T + 3) + ɛ T +3 }{{}}{{} Y T +3 T ˆɛ T +3 T The forecast is a continuation of the estimated trend line, and the forecast standard errors are constant (no matter how far into the future you project). This seems unrealistic of long-term projections of real economic time series. We use industrial production as an example. We fit deterministic and stochastic trend models to the series from Jan 1983 to Dec 2007, and project over Jan 2008 to Dec One would ordinarily not (shouldn t) make such a long term projection, but we do so to illustrate some differences between projecting with the two kinds of trend. df_mth <- read_csv("timeseries_monthly.csv") glimpse(df_mth) Observations: 420 Variables: 6 $ DATE <chr> "Jan-83", "Feb-83", "Mar-83", "Apr-83", "May-83",... $ ELEC_GEN_SG <dbl> 667.3, 586.9, 727.4, 719.0, 727.1, 728.4, 741.1, 7... $ TOUR_SG <int> , , , , , , $ IP_SG <dbl> 14.34, 11.37, 14.50, 12.86, 13.01, 12.62, 13.56, 1... $ CPI_US <dbl> 97.8, 97.9, 97.9, 98.6, 99.2, 99.5, 99.9, 100.2, 1... $ DOMEX5_SG <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA... IP_SG.xts <- xts(df_mth$ip_sg, order.by = as.yearmon(df_mth$date, "%B-%y")) IP_SG.ts <- as.ts(ip_sg.xts, start=start(ip_sg.xts)) autoplot(log(ip_sg.ts)) + ts_thm + theme(aspect.ratio = 1/2) In the following code, the raw series is fed into the Arima function, where it is put through a

8 Intermediate Econometrics / Time Series Analysis and Forecasting 8 box-cox transformation Y λ t 1 λ An application of L hopital s rule shows that this transformation approaches the natural log transformation when λ approaches zero. Setting λ = 0 in the Arima function gives the natural log transformation. When the forecast is made, the data is transformed back into the original scale. We begin with a pure deterministic trend plus seasonal dummies model. trendvar <- seq_along(ip_sg.ts) seasonalvars <- seasonaldummy(ip_sg.ts) Xreg = cbind(trendvar, seasonalvars) fit2 <- Arima(window(IP_SG.ts,start=c(1983,1), end=c(2008,12)), summary(fit2) order=c(0,0,0), xreg=xreg[1:312,], # Jan 1983 to Dec 2007 lambda=0, # box-cox transformation, lambda=0 is log-transformation biasadj = T) Series: window(ip_sg.ts, start = c(1983, 1), end = c(2008, 12)) Regression with ARIMA(0,0,0) errors Box Cox transformation: lambda= 0 Coefficients: intercept trendvar Jan Feb Mar Apr May s.e Jun Jul Aug Sep Oct Nov s.e sigma^2 estimated as : log likelihood= AIC= AICc= BIC= Training set error measures: ME RMSE MAE MPE MAPE MASE Training set ACF1 Training set fcst <- forecast(fit2, h=108, xreg=xreg[313:420,], biasadj=t) # forecast 1 to 108 steps (9 years x 12 quarters) ahead

9 Intermediate Econometrics / Time Series Analysis and Forecasting 9 fcst.plot <- ts.union(ip_sg.ts, fitted(fit2), fcst$mean, fcst$lower[,1], fcst$upper[,1]) # 80% autoplot(window(fcst.plot, start=c(2001,1))) + scale_color_manual(values=rep("black", 5)) + ylab("") + xlab("") + aes(linetype=series, size=series) + scale_size_manual(values=c(0.5,1,0.5,0.5,0.5)) + scale_linetype_manual(values=c("solid", "solid", "dashed", rep("dotted",4))) + ts_thm + theme(legend.position="none") The forecast plot is of the back-transformed series, which is why the projection is exponential rather than linear despite having fit a linear trend. We plot the actual (both in-sample and out-of-sample), the fitted (bold), the forecast (dashed), and 95% bounds (dotted). You can see that the projection is a continuation of the fitted line, and the forecast interval quickly becomes constant width, even though you are projecting many years into the future from a fixed time-point. The forecasts are obviously bad, since the series deviated from the fitted trend significantly over the forecast sample. Xreg = cbind(seasonalvars) fit3 <- Arima(window(IP_SG.ts,start=c(1983,1), end=c(2008,12)), order=c(1,1,0), xreg=xreg[1:312,], include.drift = T, lambda=0, # box-cox transformation, lambda=0 is log-transformation biasadj = T)

10 Intermediate Econometrics / Time Series Analysis and Forecasting 10 summary(fit3) Series: window(ip_sg.ts, start = c(1983, 1), end = c(2008, 12)) Regression with ARIMA(1,1,0) errors Box Cox transformation: lambda= 0 Coefficients: ar1 drift Jan Feb Mar Apr May Jun s.e Jul Aug Sep Oct Nov s.e sigma^2 estimated as : log likelihood= AIC= AICc= BIC= Training set error measures: ME RMSE MAE MPE MAPE MASE Training set ACF1 Training set fcst <- forecast(fit3, h=108, xreg=xreg[313:420,], biasadj=t) # forecast 1 to 120 steps (10 years x 12 quarters) ahead fcst.plot <- ts.union(ip_sg.ts, fitted(fit3), fcst$mean, fcst$lower[,1], fcst$upper[,1]) # 80% autoplot(window(fcst.plot, start=c(2001,1))) + scale_color_manual(values=rep("black", 5)) + ylab("") + xlab("") + aes(linetype=series, size=series) + scale_size_manual(values=c(0.5,1,0.5,0.5,0.5)) + scale_linetype_manual(values=c("solid", "solid", "dashed", rep("dotted",4))) + ts_thm + theme(legend.position="none")

11 Intermediate Econometrics / Time Series Analysis and Forecasting There are two main differences between the stochastic trend forecasts and the deterministic trend forecasts. The stochastic trend projection is ultimate merely a projection of the fitted trend (drift) but it starts from the last realization. The forecast interval increases with horizon, but this time without bound, which is much more realistic. In both cases there are periods where the long-run predictions appear to do very well. These episodes are quite coincidental. The differences between linear deterministic trend and stochastic trend forecasts are much more stark when looking at the sequence of one-step ahead forecasts, i.e., you update your sample every period, and forecast only one-step ahead each time. #Xreg = cbind(trendvar, seasonalvars) Xreg = seasonalvars fcst1step<-ts(matrix(rep(na,108*3),ncol=3), start=c(2009,1), frequency=12) # to store forecasts colnames(fcst1step) <- c("mean", "lower", "upper") for (i in 1:108){ fit1 <- Arima(IP_SG.ts[1:(312+i-1)], order=c(0,1,0), xreg=xreg[1:(312+i-1),], include.constant = T, biasadj = T, lambda=0) temp <- forecast(fit1, h=1, xreg=matrix(xreg[(312+i):(312+i),], nrow=1))

12 Intermediate Econometrics / Time Series Analysis and Forecasting 12 fcst1step[i,]<-cbind(temp$mean, temp$lower[,"80%"], temp$upper[,"80%"]) } ts.plot <- ts.union(actual=window(ip_sg.ts,start=c(2004,1)),fcst1step) autoplot(ts.plot) + scale_color_manual(values=rep("black", 4)) + ylab("") + xlab("") + aes(linetype=series) + scale_linetype_manual(values=c("solid", "dashed", rep("dotted",2))) + ts_thm + theme(legend.position="none") Xreg = cbind(trendvar, seasonalvars) fcst1step<-ts(matrix(rep(na,108*3),ncol=3), start=c(2009,1), frequency=12) # to store forecasts colnames(fcst1step) <- c("mean", "lower", "upper") for (i in 1:108){ fit1 <- Arima(IP_SG.ts[1:(312+i-1)], order=c(0,0,0), xreg=xreg[1:(312+i-1),], include.constant = T, biasadj = T, lambda=0) temp <- forecast(fit1, h=1, xreg=matrix(xreg[(312+i):(312+i),], nrow=1)) fcst1step[i,]<-cbind(temp$mean, temp$lower[,"80%"], temp$upper[,"80%"])

13 Intermediate Econometrics / Time Series Analysis and Forecasting 13 } ts.plot <- ts.union(actual=window(ip_sg.ts,start=c(2004,1)),fcst1step) autoplot(ts.plot) + scale_color_manual(values=rep("black", 4)) + ylab("") + xlab("") + aes(linetype=series) + scale_linetype_manual(values=c("solid", "dashed", rep("dotted",2))) + ts_thm + theme(legend.position="none") The one-step ahead forecasts from a deterministic time trend model is essentially the same as the long-range forecast, since it merely follows the fitted trend line. Finally, we end off with forecasts with models selected from the forecast package auto.arima. In our implementation, we allow the algorithm to select a new model each period. The algorithm decides (via KPSS) whether or not to difference the series (both first and seasonal differencing), and then to choose the order (both non-seasonal and seasonal ARMA orders). In the code below, there is a line that you can un-comment (print(fit1$arma)) to ask the algorithm to display the model choice (ar, ma seasonal ar, seasonal ma, frequency, no of differencing, no of seasonal differencing). There is some variation at first, but it eventually settles down to ARMA(0, 1, 1)(0, 0, 2) 4. fcst1step<-ts(matrix(rep(na,108*3),ncol=3), start=c(2009,1), frequency=12) # to store forecasts colnames(fcst1step) <- c("mean", "lower", "upper") for (i in 1:108){

14 Intermediate Econometrics / Time Series Analysis and Forecasting 14 fit1 <- auto.arima(ts(ip_sg.ts[1:(312+i-1)], start=c(1983,1), frequency=12), lambda=0, biasadj = TRUE, seasonal = T) # print(fit1$arma) temp <- forecast(fit1, h=1) fcst1step[i,]<-cbind(temp$mean, temp$lower[,"80%"], temp$upper[,"80%"]) } ts.plot <- ts.union(actual=window(ip_sg.ts,start=c(2004,1)),fcst1step) autoplot(ts.plot) + scale_color_manual(values=rep("black", 4)) + ylab("") + xlab("") + aes(linetype=series) + scale_linetype_manual(values=c("solid", "dashed", rep("dotted",2))) + ts_thm + theme(legend.position="none") Exercises 1. Show that the forecasts Y T +h T from a stationary AR(1) with β 1 (0, 1) decreases monotonically towards the unconditional mean of the AR(1) if Y T is above it, and increases monotonically towards the unconditional mean of the AR(1) if Y T is below it. (Hint: we have already shown convergence to the mean; check the direction of convergence in each case by deriving an appropriate expression for Y T +h+1 T Y T +h T ).

Transformations for variance stabilization

Transformations for variance stabilization FORECASTING USING R Transformations for variance stabilization Rob Hyndman Author, forecast Variance stabilization If the data show increasing variation as the level of the series increases, then a transformation

More information

Introduction to Time Series Modelling This version: 29 November 2018

Introduction to Time Series Modelling This version: 29 November 2018 Introduction to Time Series Modelling This version: 29 November 2018 Notes for Intermediate Econometrics / Time Series Analysis and Forecasting Anthony Tay # This note uses the following libraries library(tidyverse)

More information

Forecasting using R. Rob J Hyndman. 2.4 Non-seasonal ARIMA models. Forecasting using R 1

Forecasting using R. Rob J Hyndman. 2.4 Non-seasonal ARIMA models. Forecasting using R 1 Forecasting using R Rob J Hyndman 2.4 Non-seasonal ARIMA models Forecasting using R 1 Outline 1 Autoregressive models 2 Moving average models 3 Non-seasonal ARIMA models 4 Partial autocorrelations 5 Estimation

More information

477/577 In-class Exercise 5 : Fitting Wine Sales

477/577 In-class Exercise 5 : Fitting Wine Sales 477/577 In-class Exercise 5 : Fitting Wine Sales (due Fri 4/06/2017) Name: Use this file as a template for your report. Submit your code and comments together with (selected) output from R console. Your

More information

Suan Sunandha Rajabhat University

Suan Sunandha Rajabhat University Forecasting Exchange Rate between Thai Baht and the US Dollar Using Time Series Analysis Kunya Bowornchockchai Suan Sunandha Rajabhat University INTRODUCTION The objective of this research is to forecast

More information

TIME SERIES ANALYSIS AND FORECASTING USING THE STATISTICAL MODEL ARIMA

TIME SERIES ANALYSIS AND FORECASTING USING THE STATISTICAL MODEL ARIMA CHAPTER 6 TIME SERIES ANALYSIS AND FORECASTING USING THE STATISTICAL MODEL ARIMA 6.1. Introduction A time series is a sequence of observations ordered in time. A basic assumption in the time series analysis

More information

GAMINGRE 8/1/ of 7

GAMINGRE 8/1/ of 7 FYE 09/30/92 JULY 92 0.00 254,550.00 0.00 0 0 0 0 0 0 0 0 0 254,550.00 0.00 0.00 0.00 0.00 254,550.00 AUG 10,616,710.31 5,299.95 845,656.83 84,565.68 61,084.86 23,480.82 339,734.73 135,893.89 67,946.95

More information

Homework 5, Problem 1 Andrii Baryshpolets 6 April 2017

Homework 5, Problem 1 Andrii Baryshpolets 6 April 2017 Homework 5, Problem 1 Andrii Baryshpolets 6 April 2017 Total Private Residential Construction Spending library(quandl) Warning: package 'Quandl' was built under R version 3.3.3 Loading required package:

More information

FORECASTING USING R. Dynamic regression. Rob Hyndman Author, forecast

FORECASTING USING R. Dynamic regression. Rob Hyndman Author, forecast FORECASTING USING R Dynamic regression Rob Hyndman Author, forecast Dynamic regression Regression model with ARIMA errors y t = β 0 + β 1 x 1,t + + β r x r,t + e t y t modeled as function of r explanatory

More information

STAT 436 / Lecture 16: Key

STAT 436 / Lecture 16: Key STAT 436 / 536 - Lecture 16: Key Modeling Non-Stationary Time Series Many time series models are non-stationary. Recall a time series is stationary if the mean and variance are constant in time and the

More information

EXAMINATIONS OF THE ROYAL STATISTICAL SOCIETY

EXAMINATIONS OF THE ROYAL STATISTICAL SOCIETY EXAMINATIONS OF THE ROYAL STATISTICAL SOCIETY GRADUATE DIPLOMA, 011 MODULE 3 : Stochastic processes and time series Time allowed: Three Hours Candidates should answer FIVE questions. All questions carry

More information

Technical note on seasonal adjustment for M0

Technical note on seasonal adjustment for M0 Technical note on seasonal adjustment for M0 July 1, 2013 Contents 1 M0 2 2 Steps in the seasonal adjustment procedure 3 2.1 Pre-adjustment analysis............................... 3 2.2 Seasonal adjustment.................................

More information

Forecasting using R. Rob J Hyndman. 3.2 Dynamic regression. Forecasting using R 1

Forecasting using R. Rob J Hyndman. 3.2 Dynamic regression. Forecasting using R 1 Forecasting using R Rob J Hyndman 3.2 Dynamic regression Forecasting using R 1 Outline 1 Regression with ARIMA errors 2 Stochastic and deterministic trends 3 Periodic seasonality 4 Lab session 14 5 Dynamic

More information

Case Study: Modelling Industrial Dryer Temperature Arun K. Tangirala 11/19/2016

Case Study: Modelling Industrial Dryer Temperature Arun K. Tangirala 11/19/2016 Case Study: Modelling Industrial Dryer Temperature Arun K. Tangirala 11/19/2016 Background This is a case study concerning time-series modelling of the temperature of an industrial dryer. Data set contains

More information

Time Series in R: Forecasting and Visualisation. Forecast evaluation 29 May 2017

Time Series in R: Forecasting and Visualisation. Forecast evaluation 29 May 2017 Time Series in R: Forecasting and Visualisation Forecast evaluation 29 May 2017 1 Outline 1 Forecasting residuals 2 Evaluating forecast accuracy 3 Forecasting benchmark methods 4 Lab session 7 5 Time series

More information

Data %>% Power %>% R You ready?

Data %>% Power %>% R You ready? Data %>% Power %>% R You ready? R a powerful open source programming language available in the database as Oracle R Enterprise that lets you do virtually anything Especially (but not only) Statistics Machine

More information

Technical note on seasonal adjustment for Capital goods imports

Technical note on seasonal adjustment for Capital goods imports Technical note on seasonal adjustment for Capital goods imports July 1, 2013 Contents 1 Capital goods imports 2 1.1 Additive versus multiplicative seasonality..................... 2 2 Steps in the seasonal

More information

Time Series Analysis

Time Series Analysis Time Series Analysis A time series is a sequence of observations made: 1) over a continuous time interval, 2) of successive measurements across that interval, 3) using equal spacing between consecutive

More information

BUSI 460 Suggested Answers to Selected Review and Discussion Questions Lesson 7

BUSI 460 Suggested Answers to Selected Review and Discussion Questions Lesson 7 BUSI 460 Suggested Answers to Selected Review and Discussion Questions Lesson 7 1. The definitions follow: (a) Time series: Time series data, also known as a data series, consists of observations on a

More information

INTRODUCTION TO TIME SERIES ANALYSIS. The Simple Moving Average Model

INTRODUCTION TO TIME SERIES ANALYSIS. The Simple Moving Average Model INTRODUCTION TO TIME SERIES ANALYSIS The Simple Moving Average Model The Simple Moving Average Model The simple moving average (MA) model: More formally: where t is mean zero white noise (WN). Three parameters:

More information

Dummy Variables. Susan Thomas IGIDR, Bombay. 24 November, 2008

Dummy Variables. Susan Thomas IGIDR, Bombay. 24 November, 2008 IGIDR, Bombay 24 November, 2008 The problem of structural change Model: Y i = β 0 + β 1 X 1i + ɛ i Structural change, type 1: change in parameters in time. Y i = α 1 + β 1 X i + e 1i for period 1 Y i =

More information

Classical Decomposition Model Revisited: I

Classical Decomposition Model Revisited: I Classical Decomposition Model Revisited: I recall classical decomposition model for time series Y t, namely, Y t = m t + s t + W t, where m t is trend; s t is periodic with known period s (i.e., s t s

More information

Univariate ARIMA Models

Univariate ARIMA Models Univariate ARIMA Models ARIMA Model Building Steps: Identification: Using graphs, statistics, ACFs and PACFs, transformations, etc. to achieve stationary and tentatively identify patterns and model components.

More information

FORECASTING COARSE RICE PRICES IN BANGLADESH

FORECASTING COARSE RICE PRICES IN BANGLADESH Progress. Agric. 22(1 & 2): 193 201, 2011 ISSN 1017-8139 FORECASTING COARSE RICE PRICES IN BANGLADESH M. F. Hassan*, M. A. Islam 1, M. F. Imam 2 and S. M. Sayem 3 Department of Agricultural Statistics,

More information

Forecasting using R. Rob J Hyndman. 2.5 Seasonal ARIMA models. Forecasting using R 1

Forecasting using R. Rob J Hyndman. 2.5 Seasonal ARIMA models. Forecasting using R 1 Forecasting using R Rob J Hyndman 2.5 Seasonal ARIMA models Forecasting using R 1 Outline 1 Backshift notation reviewed 2 Seasonal ARIMA models 3 ARIMA vs ETS 4 Lab session 12 Forecasting using R Backshift

More information

Time Series Analysis of United States of America Crude Oil and Petroleum Products Importations from Saudi Arabia

Time Series Analysis of United States of America Crude Oil and Petroleum Products Importations from Saudi Arabia International Journal of Applied Science and Technology Vol. 5, No. 5; October 2015 Time Series Analysis of United States of America Crude Oil and Petroleum Products Importations from Saudi Arabia Olayan

More information

arxiv: v1 [stat.co] 11 Dec 2012

arxiv: v1 [stat.co] 11 Dec 2012 Simulating the Continuation of a Time Series in R December 12, 2012 arxiv:1212.2393v1 [stat.co] 11 Dec 2012 Halis Sak 1 Department of Industrial and Systems Engineering, Yeditepe University, Kayışdağı,

More information

Lecture 8: ARIMA Forecasting Please read Chapters 7 and 8 of MWH Book

Lecture 8: ARIMA Forecasting Please read Chapters 7 and 8 of MWH Book Lecture 8: ARIMA Forecasting Please read Chapters 7 and 8 of MWH Book 1 Predicting Error 1. y denotes a random variable (stock price, weather, etc) 2. Sometimes we want to do prediction (guessing). Let

More information

Testing for non-stationarity

Testing for non-stationarity 20 November, 2009 Overview The tests for investigating the non-stationary of a time series falls into four types: 1 Check the null that there is a unit root against stationarity. Within these, there are

More information

Forecasting using R. Rob J Hyndman. 1.3 Seasonality and trends. Forecasting using R 1

Forecasting using R. Rob J Hyndman. 1.3 Seasonality and trends. Forecasting using R 1 Forecasting using R Rob J Hyndman 1.3 Seasonality and trends Forecasting using R 1 Outline 1 Time series components 2 STL decomposition 3 Forecasting and decomposition 4 Lab session 5 Forecasting using

More information

The log transformation produces a time series whose variance can be treated as constant over time.

The log transformation produces a time series whose variance can be treated as constant over time. TAT 520 Homework 6 Fall 2017 Note: Problem 5 is mandatory for graduate students and extra credit for undergraduates. 1) The quarterly earnings per share for 1960-1980 are in the object in the TA package.

More information

Summary of Seasonal Normal Review Investigations CWV Review

Summary of Seasonal Normal Review Investigations CWV Review Summary of Seasonal Normal Review Investigations CWV Review DESC 31 st March 2009 1 Contents Stage 1: The Composite Weather Variable (CWV) An Introduction / background Understanding of calculation Stage

More information

Lecture 2: Univariate Time Series

Lecture 2: Univariate Time Series Lecture 2: Univariate Time Series Analysis: Conditional and Unconditional Densities, Stationarity, ARMA Processes Prof. Massimo Guidolin 20192 Financial Econometrics Spring/Winter 2017 Overview Motivation:

More information

Time Series and Forecasting Using R

Time Series and Forecasting Using R Time Series and Forecasting Using R Lab Sessions 25 May 2016 Before doing any exercises in R, load the fpp package using library(fpp) and the ggplot2 package using library(ggplot2). Lab Session: Time series

More information

Applied Forecasting (LECTURENOTES) Prof. Rozenn Dahyot

Applied Forecasting (LECTURENOTES) Prof. Rozenn Dahyot Applied Forecasting (LECTURENOTES) Prof. Rozenn Dahyot SCHOOL OF COMPUTER SCIENCE AND STATISTICS TRINITY COLLEGE DUBLIN IRELAND https://www.scss.tcd.ie/rozenn.dahyot Michaelmas Term 2017 Contents 1 Introduction

More information

Time series and Forecasting

Time series and Forecasting Chapter 2 Time series and Forecasting 2.1 Introduction Data are frequently recorded at regular time intervals, for instance, daily stock market indices, the monthly rate of inflation or annual profit figures.

More information

Ch 5. Models for Nonstationary Time Series. Time Series Analysis

Ch 5. Models for Nonstationary Time Series. Time Series Analysis We have studied some deterministic and some stationary trend models. However, many time series data cannot be modeled in either way. Ex. The data set oil.price displays an increasing variation from the

More information

Robust Long Term Forecasting of Seasonal Time Series

Robust Long Term Forecasting of Seasonal Time Series Robust Long Term Forecasting of Seasonal Time Series Jörg Wichard 1 and Christian Merkwirth 2 1 Institute of Molecular Pharmacology, Robert-Rössle-Str. 10, D-13125 Berlin, Germany JoergWichard@web.de,

More information

ECON/FIN 250: Forecasting in Finance and Economics: Section 7: Unit Roots & Dickey-Fuller Tests

ECON/FIN 250: Forecasting in Finance and Economics: Section 7: Unit Roots & Dickey-Fuller Tests ECON/FIN 250: Forecasting in Finance and Economics: Section 7: Unit Roots & Dickey-Fuller Tests Patrick Herb Brandeis University Spring 2016 Patrick Herb (Brandeis University) Unit Root Tests ECON/FIN

More information

STATISTICS OF CLIMATE EXTREMES// TRENDS IN CLIMATE DATASETS

STATISTICS OF CLIMATE EXTREMES// TRENDS IN CLIMATE DATASETS STATISTICS OF CLIMATE EXTREMES// TRENDS IN CLIMATE DATASETS Richard L Smith Departments of STOR and Biostatistics, University of North Carolina at Chapel Hill and Statistical and Applied Mathematical Sciences

More information

Econometrics. Week 11. Fall Institute of Economic Studies Faculty of Social Sciences Charles University in Prague

Econometrics. Week 11. Fall Institute of Economic Studies Faculty of Social Sciences Charles University in Prague Econometrics Week 11 Institute of Economic Studies Faculty of Social Sciences Charles University in Prague Fall 2012 1 / 30 Recommended Reading For the today Advanced Time Series Topics Selected topics

More information

Nonparametric time series forecasting with dynamic updating

Nonparametric time series forecasting with dynamic updating 18 th World IMAS/MODSIM Congress, Cairns, Australia 13-17 July 2009 http://mssanz.org.au/modsim09 1 Nonparametric time series forecasting with dynamic updating Han Lin Shang and Rob. J. Hyndman Department

More information

Non-Stationary Time Series and Unit Root Testing

Non-Stationary Time Series and Unit Root Testing Econometrics II Non-Stationary Time Series and Unit Root Testing Morten Nyboe Tabor Course Outline: Non-Stationary Time Series and Unit Root Testing 1 Stationarity and Deviation from Stationarity Trend-Stationarity

More information

Time-Series analysis for wind speed forecasting

Time-Series analysis for wind speed forecasting Malaya Journal of Matematik, Vol. S, No. 1, 55-61, 2018 https://doi.org/10.26637/mjm0s01/11 Time-Series analysis for wind speed forecasting Garima Jain1 Abstract In this paper, an enormous amount of study

More information

Time Series Forecasting Methods:

Time Series Forecasting Methods: Corso di TRASPORTI E TERRITORIO Time Series Forecasting Methods: EXPONENTIAL SMOOTHING DOCENTI Agostino Nuzzolo (nuzzolo@ing.uniroma2.it) Antonio Comi (comi@ing.uniroma2.it) 1 Classificazione dei metodi

More information

Week 9: An Introduction to Time Series

Week 9: An Introduction to Time Series BUS41100 Applied Regression Analysis Week 9: An Introduction to Time Series Dependent data, autocorrelation, AR and periodic regression models Max H. Farrell The University of Chicago Booth School of Business

More information

Unit root problem, solution of difference equations Simple deterministic model, question of unit root

Unit root problem, solution of difference equations Simple deterministic model, question of unit root Unit root problem, solution of difference equations Simple deterministic model, question of unit root (1 φ 1 L)X t = µ, Solution X t φ 1 X t 1 = µ X t = A + Bz t with unknown z and unknown A (clearly X

More information

Determine the trend for time series data

Determine the trend for time series data Extra Online Questions Determine the trend for time series data Covers AS 90641 (Statistics and Modelling 3.1) Scholarship Statistics and Modelling Chapter 1 Essent ial exam notes Time series 1. The value

More information

Available online at ScienceDirect. Procedia Computer Science 72 (2015 )

Available online at  ScienceDirect. Procedia Computer Science 72 (2015 ) Available online at www.sciencedirect.com ScienceDirect Procedia Computer Science 72 (2015 ) 630 637 The Third Information Systems International Conference Performance Comparisons Between Arima and Arimax

More information

Introduction to S+FinMetrics

Introduction to S+FinMetrics 2008 An Introduction to S+FinMetrics This workshop will unleash the power of S+FinMetrics module for the econometric modeling and prediction of economic and financial time series. S+FinMetrics library

More information

Annual Average NYMEX Strip Comparison 7/03/2017

Annual Average NYMEX Strip Comparison 7/03/2017 Annual Average NYMEX Strip Comparison 7/03/2017 To Year to Year Oil Price Deck ($/bbl) change Year change 7/3/2017 6/1/2017 5/1/2017 4/3/2017 3/1/2017 2/1/2017-2.7% 2017 Average -10.4% 47.52 48.84 49.58

More information

Approximating Fixed-Horizon Forecasts Using Fixed-Event Forecasts

Approximating Fixed-Horizon Forecasts Using Fixed-Event Forecasts Approximating Fixed-Horizon Forecasts Using Fixed-Event Forecasts Malte Knüppel and Andreea L. Vladu Deutsche Bundesbank 9th ECB Workshop on Forecasting Techniques 4 June 216 This work represents the authors

More information

Reliability and Risk Analysis. Time Series, Types of Trend Functions and Estimates of Trends

Reliability and Risk Analysis. Time Series, Types of Trend Functions and Estimates of Trends Reliability and Risk Analysis Stochastic process The sequence of random variables {Y t, t = 0, ±1, ±2 } is called the stochastic process The mean function of a stochastic process {Y t} is the function

More information

Chapter 9: Forecasting

Chapter 9: Forecasting Chapter 9: Forecasting One of the critical goals of time series analysis is to forecast (predict) the values of the time series at times in the future. When forecasting, we ideally should evaluate the

More information

Lecture Prepared By: Mohammad Kamrul Arefin Lecturer, School of Business, North South University

Lecture Prepared By: Mohammad Kamrul Arefin Lecturer, School of Business, North South University Lecture 15 20 Prepared By: Mohammad Kamrul Arefin Lecturer, School of Business, North South University Modeling for Time Series Forecasting Forecasting is a necessary input to planning, whether in business,

More information

Empirical Market Microstructure Analysis (EMMA)

Empirical Market Microstructure Analysis (EMMA) Empirical Market Microstructure Analysis (EMMA) Lecture 3: Statistical Building Blocks and Econometric Basics Prof. Dr. Michael Stein michael.stein@vwl.uni-freiburg.de Albert-Ludwigs-University of Freiburg

More information

at least 50 and preferably 100 observations should be available to build a proper model

at least 50 and preferably 100 observations should be available to build a proper model III Box-Jenkins Methods 1. Pros and Cons of ARIMA Forecasting a) need for data at least 50 and preferably 100 observations should be available to build a proper model used most frequently for hourly or

More information

NATCOR Regression Modelling for Time Series

NATCOR Regression Modelling for Time Series Universität Hamburg Institut für Wirtschaftsinformatik Prof. Dr. D.B. Preßmar Professor Robert Fildes NATCOR Regression Modelling for Time Series The material presented has been developed with the substantial

More information

Multiple Regression Analysis

Multiple Regression Analysis 1 OUTLINE Analysis of Data and Model Hypothesis Testing Dummy Variables Research in Finance 2 ANALYSIS: Types of Data Time Series data Cross-Sectional data Panel data Trend Seasonal Variation Cyclical

More information

Package TimeSeries.OBeu

Package TimeSeries.OBeu Type Package Package TimeSeries.OBeu Title Time Series Analysis 'OpenBudgets' Version 1.2.2 Date 2018-01-20 January 22, 2018 Estimate and return the needed parameters for visualisations designed for 'OpenBudgets'

More information

DESC Technical Workgroup. CWV Optimisation Production Phase Results. 17 th November 2014

DESC Technical Workgroup. CWV Optimisation Production Phase Results. 17 th November 2014 DESC Technical Workgroup CWV Optimisation Production Phase Results 17 th November 2014 1 2 Contents CWV Optimisation Background Trial Phase Production Phase Explanation of Results Production Phase Results

More information

Dynamic linear models (aka state-space models) 1

Dynamic linear models (aka state-space models) 1 Dynamic linear models (aka state-space models) 1 Advanced Econometris: Time Series Hedibert Freitas Lopes INSPER 1 Part of this lecture is based on Gamerman and Lopes (2006) Markov Chain Monte Carlo: Stochastic

More information

Forecasting Gold Price. A Comparative Study

Forecasting Gold Price. A Comparative Study Course of Financial Econometrics FIRM Forecasting Gold Price A Comparative Study Alessio Azzutti, University of Florence Abstract: This paper seeks to evaluate the appropriateness of a variety of existing

More information

A Dynamic-Trend Exponential Smoothing Model

A Dynamic-Trend Exponential Smoothing Model City University of New York (CUNY) CUNY Academic Works Publications and Research Baruch College Summer 2007 A Dynamic-Trend Exponential Smoothing Model Don Miller Virginia Commonwealth University Dan Williams

More information

Part 1. Multiple Choice (50 questions, 1 point each) Part 2. Problems/Short Answer (10 questions, 5 points each)

Part 1. Multiple Choice (50 questions, 1 point each) Part 2. Problems/Short Answer (10 questions, 5 points each) GROUND RULES: This exam contains two parts: Part 1. Multiple Choice (50 questions, 1 point each) Part 2. Problems/Short Answer (10 questions, 5 points each) The maximum number of points on this exam is

More information

Multiple Regression Analysis

Multiple Regression Analysis 1 OUTLINE Basic Concept: Multiple Regression MULTICOLLINEARITY AUTOCORRELATION HETEROSCEDASTICITY REASEARCH IN FINANCE 2 BASIC CONCEPTS: Multiple Regression Y i = β 1 + β 2 X 1i + β 3 X 2i + β 4 X 3i +

More information

AR, MA and ARMA models

AR, MA and ARMA models AR, MA and AR by Hedibert Lopes P Based on Tsay s Analysis of Financial Time Series (3rd edition) P 1 Stationarity 2 3 4 5 6 7 P 8 9 10 11 Outline P Linear Time Series Analysis and Its Applications For

More information

Advanced Econometrics

Advanced Econometrics Advanced Econometrics Marco Sunder Nov 04 2010 Marco Sunder Advanced Econometrics 1/ 25 Contents 1 2 3 Marco Sunder Advanced Econometrics 2/ 25 Music Marco Sunder Advanced Econometrics 3/ 25 Music Marco

More information

Non-Stationary Time Series and Unit Root Testing

Non-Stationary Time Series and Unit Root Testing Econometrics II Non-Stationary Time Series and Unit Root Testing Morten Nyboe Tabor Course Outline: Non-Stationary Time Series and Unit Root Testing 1 Stationarity and Deviation from Stationarity Trend-Stationarity

More information

Report for Fitting a Time Series Model

Report for Fitting a Time Series Model Report for Fitting a Time Series Model 1. Introduction In this report, I choose the stock price of Starbucks Corporation (SBUX) as my data to analyze. The time period for the data is from Jan 2009 to Dec

More information

Design of a Weather-Normalization Forecasting Model

Design of a Weather-Normalization Forecasting Model Design of a Weather-Normalization Forecasting Model Final Briefing 09 May 2014 Sponsor: Northern Virginia Electric Cooperative Abram Gross Jedidiah Shirey Yafeng Peng OR-699 Agenda Background Problem Statement

More information

Problem Set 2 Solution Sketches Time Series Analysis Spring 2010

Problem Set 2 Solution Sketches Time Series Analysis Spring 2010 Problem Set 2 Solution Sketches Time Series Analysis Spring 2010 Forecasting 1. Let X and Y be two random variables such that E(X 2 ) < and E(Y 2 )

More information

4. MA(2) +drift: y t = µ + ɛ t + θ 1 ɛ t 1 + θ 2 ɛ t 2. Mean: where θ(l) = 1 + θ 1 L + θ 2 L 2. Therefore,

4. MA(2) +drift: y t = µ + ɛ t + θ 1 ɛ t 1 + θ 2 ɛ t 2. Mean: where θ(l) = 1 + θ 1 L + θ 2 L 2. Therefore, 61 4. MA(2) +drift: y t = µ + ɛ t + θ 1 ɛ t 1 + θ 2 ɛ t 2 Mean: y t = µ + θ(l)ɛ t, where θ(l) = 1 + θ 1 L + θ 2 L 2. Therefore, E(y t ) = µ + θ(l)e(ɛ t ) = µ 62 Example: MA(q) Model: y t = ɛ t + θ 1 ɛ

More information

Comparing the Univariate Modeling Techniques, Box-Jenkins and Artificial Neural Network (ANN) for Measuring of Climate Index

Comparing the Univariate Modeling Techniques, Box-Jenkins and Artificial Neural Network (ANN) for Measuring of Climate Index Applied Mathematical Sciences, Vol. 8, 2014, no. 32, 1557-1568 HIKARI Ltd, www.m-hikari.com http://dx.doi.org/10.12988/ams.2014.4150 Comparing the Univariate Modeling Techniques, Box-Jenkins and Artificial

More information

Non-Stationary Time Series and Unit Root Testing

Non-Stationary Time Series and Unit Root Testing Econometrics II Non-Stationary Time Series and Unit Root Testing Morten Nyboe Tabor Course Outline: Non-Stationary Time Series and Unit Root Testing 1 Stationarity and Deviation from Stationarity Trend-Stationarity

More information

Marcel Dettling. Applied Time Series Analysis FS 2012 Week 01. ETH Zürich, February 20, Institute for Data Analysis and Process Design

Marcel Dettling. Applied Time Series Analysis FS 2012 Week 01. ETH Zürich, February 20, Institute for Data Analysis and Process Design Marcel Dettling Institute for Data Analysis and Process Design Zurich University of Applied Sciences marcel.dettling@zhaw.ch http://stat.ethz.ch/~dettling ETH Zürich, February 20, 2012 1 Your Lecturer

More information

Lecture 6a: Unit Root and ARIMA Models

Lecture 6a: Unit Root and ARIMA Models Lecture 6a: Unit Root and ARIMA Models 1 2 Big Picture A time series is non-stationary if it contains a unit root unit root nonstationary The reverse is not true. For example, y t = cos(t) + u t has no

More information

Time Series I Time Domain Methods

Time Series I Time Domain Methods Astrostatistics Summer School Penn State University University Park, PA 16802 May 21, 2007 Overview Filtering and the Likelihood Function Time series is the study of data consisting of a sequence of DEPENDENT

More information

Nonstationary Time Series:

Nonstationary Time Series: Nonstationary Time Series: Unit Roots Egon Zakrajšek Division of Monetary Affairs Federal Reserve Board Summer School in Financial Mathematics Faculty of Mathematics & Physics University of Ljubljana September

More information

Prashant Pant 1, Achal Garg 2 1,2 Engineer, Keppel Offshore and Marine Engineering India Pvt. Ltd, Mumbai. IJRASET 2013: All Rights are Reserved 356

Prashant Pant 1, Achal Garg 2 1,2 Engineer, Keppel Offshore and Marine Engineering India Pvt. Ltd, Mumbai. IJRASET 2013: All Rights are Reserved 356 Forecasting Of Short Term Wind Power Using ARIMA Method Prashant Pant 1, Achal Garg 2 1,2 Engineer, Keppel Offshore and Marine Engineering India Pvt. Ltd, Mumbai Abstract- Wind power, i.e., electrical

More information

Time Series and Forecasting

Time Series and Forecasting Time Series and Forecasting Introduction to Forecasting n What is forecasting? n Primary Function is to Predict the Future using (time series related or other) data we have in hand n Why are we interested?

More information

Stat 153 Time Series. Problem Set 4

Stat 153 Time Series. Problem Set 4 Stat Time Series Problem Set Problem I generated 000 observations from the MA() model with parameter 0.7 using the following R function and then fitted the ARMA(, ) model to the data. order = c(, 0, )),

More information

Multivariate Regression Model Results

Multivariate Regression Model Results Updated: August, 0 Page of Multivariate Regression Model Results 4 5 6 7 8 This exhibit provides the results of the load model forecast discussed in Schedule. Included is the forecast of short term system

More information

Introduction to Forecasting

Introduction to Forecasting Introduction to Forecasting Introduction to Forecasting Predicting the future Not an exact science but instead consists of a set of statistical tools and techniques that are supported by human judgment

More information

Modelling Seasonality of Gross Domestic Product in Belgium

Modelling Seasonality of Gross Domestic Product in Belgium University of Vienna Univ.-Prof. Dipl.-Ing. Dr. Robert M. Kunst Course: Econometric Analysis of Seasonal Time Series (Oekonometrie der Saison) Working Paper Modelling Seasonality of Gross Domestic Product

More information

University of Oxford. Statistical Methods Autocorrelation. Identification and Estimation

University of Oxford. Statistical Methods Autocorrelation. Identification and Estimation University of Oxford Statistical Methods Autocorrelation Identification and Estimation Dr. Órlaith Burke Michaelmas Term, 2011 Department of Statistics, 1 South Parks Road, Oxford OX1 3TG Contents 1 Model

More information

Lecture Prepared By: Mohammad Kamrul Arefin Lecturer, School of Business, North South University

Lecture Prepared By: Mohammad Kamrul Arefin Lecturer, School of Business, North South University Lecture 15 20 Prepared By: Mohammad Kamrul Arefin Lecturer, School of Business, North South University Modeling for Time Series Forecasting Forecasting is a necessary input to planning, whether in business,

More information

Forecasting Area, Production and Yield of Cotton in India using ARIMA Model

Forecasting Area, Production and Yield of Cotton in India using ARIMA Model Forecasting Area, Production and Yield of Cotton in India using ARIMA Model M. K. Debnath 1, Kartic Bera 2 *, P. Mishra 1 1 Department of Agricultural Statistics, Bidhan Chanda Krishi Vishwavidyalaya,

More information

A time series is called strictly stationary if the joint distribution of every collection (Y t

A time series is called strictly stationary if the joint distribution of every collection (Y t 5 Time series A time series is a set of observations recorded over time. You can think for example at the GDP of a country over the years (or quarters) or the hourly measurements of temperature over a

More information

AR(p) + I(d) + MA(q) = ARIMA(p, d, q)

AR(p) + I(d) + MA(q) = ARIMA(p, d, q) AR(p) + I(d) + MA(q) = ARIMA(p, d, q) Outline 1 4.1: Nonstationarity in the Mean 2 ARIMA Arthur Berg AR(p) + I(d)+ MA(q) = ARIMA(p, d, q) 2/ 19 Deterministic Trend Models Polynomial Trend Consider the

More information

Jayalath Ekanayake Jonas Tappolet Harald Gall Abraham Bernstein. Time variance and defect prediction in software projects: additional figures

Jayalath Ekanayake Jonas Tappolet Harald Gall Abraham Bernstein. Time variance and defect prediction in software projects: additional figures Jayalath Ekanayake Jonas Tappolet Harald Gall Abraham Bernstein TECHNICAL REPORT No. IFI-2.4 Time variance and defect prediction in software projects: additional figures 2 University of Zurich Department

More information

Package TSPred. April 5, 2017

Package TSPred. April 5, 2017 Type Package Package TSPred April 5, 2017 Title Functions for Benchmarking Time Series Prediction Version 3.0.2 Date 2017-04-05 Author Rebecca Pontes Salles [aut, cre, cph] (CEFET/RJ), Eduardo Ogasawara

More information

STA 6857 ARIMA and SARIMA Models ( 3.8 and 3.9)

STA 6857 ARIMA and SARIMA Models ( 3.8 and 3.9) STA 6857 ARIMA and SARIMA Models ( 3.8 and 3.9) Outline 1 Building ARIMA Models 2 SARIMA 3 Homework 4c Arthur Berg STA 6857 ARIMA and SARIMA Models ( 3.8 and 3.9) 2/ 34 Outline 1 Building ARIMA Models

More information

Computing & Telecommunications Services Monthly Report January CaTS Help Desk. Wright State University (937)

Computing & Telecommunications Services Monthly Report January CaTS Help Desk. Wright State University (937) January 215 Monthly Report Computing & Telecommunications Services Monthly Report January 215 CaTS Help Desk (937) 775-4827 1-888-775-4827 25 Library Annex helpdesk@wright.edu www.wright.edu/cats/ Last

More information

Nonstationary time series models

Nonstationary time series models 13 November, 2009 Goals Trends in economic data. Alternative models of time series trends: deterministic trend, and stochastic trend. Comparison of deterministic and stochastic trend models The statistical

More information

Applied Time Series Notes ( 1) Dates. ñ Internal: # days since Jan 1, ñ Need format for reading, one for writing

Applied Time Series Notes ( 1) Dates. ñ Internal: # days since Jan 1, ñ Need format for reading, one for writing Applied Time Series Notes ( 1) Dates ñ Internal: # days since Jan 1, 1960 ñ Need format for reading, one for writing ñ Often DATE is ID variable (extrapolates) ñ Program has lots of examples: options ls=76

More information

Decision 411: Class 7

Decision 411: Class 7 Decision 411: Class 7 Confidence limits for sums of coefficients Use of the time index as a regressor The difficulty of predicting the future Confidence intervals for sums of coefficients Sometimes the

More information

STA 6857 ARIMA and SARIMA Models ( 3.8 and 3.9) Outline. Return Rate. US Gross National Product

STA 6857 ARIMA and SARIMA Models ( 3.8 and 3.9) Outline. Return Rate. US Gross National Product STA 6857 ARIMA and SARIMA Models ( 3.8 and 3.9) Outline 1 Building ARIMA Models 2 SARIMA 3 Homework 4c Arthur Berg STA 6857 ARIMA and SARIMA Models ( 3.8 and 3.9) 2/ 34 Return Rate Suppose x t is the value

More information

FORECASTING SUGARCANE PRODUCTION IN INDIA WITH ARIMA MODEL

FORECASTING SUGARCANE PRODUCTION IN INDIA WITH ARIMA MODEL FORECASTING SUGARCANE PRODUCTION IN INDIA WITH ARIMA MODEL B. N. MANDAL Abstract: Yearly sugarcane production data for the period of - to - of India were analyzed by time-series methods. Autocorrelation

More information

Univariate, Nonstationary Processes

Univariate, Nonstationary Processes Univariate, Nonstationary Processes Jamie Monogan University of Georgia March 20, 2018 Jamie Monogan (UGA) Univariate, Nonstationary Processes March 20, 2018 1 / 14 Objectives By the end of this meeting,

More information