February 6th, 2017

Chapter 1 Questions

Is the goal of this study descriptive or predictive?

The focus of the study is to evaluate the impact of the September 11th terrorist attacks on transportation. Since it is trying to explain or understand the patterns that took place, the goal of the study is descriptive.

What is the forecast horizon to consider in this task? Are next month’s forecasts sufficient?

Since the goal is to understand the impacts of 9/11, next month’s forecast would not work as part of this study’s analysis. I would think that you would want to analyze a larger portion of data prior to the attacks, such as five to ten years. This would allow you to account for seasonality, noise, etc. You would be able to review major trends over time. On the other hand of the attacks, I would expect that the same amount of time, if not more would need to be analyzed. To be expected, post-attack would show an incredible impact on travel. However, as time went on, it would make sense that the trend would slowly return to a normal distribution.

What level of automation does this forecasting task require? Consider the four questions related to automation.

The level of automation required is little to none. The analysis uses fixed periods so it will not need to be refreshed, like a roll forward forecast would need. Four questions of automation: 1 - Three series need to be forecasted. 2 - It is a one time event that is being reviewed. 3 - Data on three series is given. 4 - Need to be able to use Excel/R Studio to process data and analyze.

What is the meaning of t=1,2,3 in the Air series? Which time period does t=1 refer to?

t=1,2,3 refers to the time periods of interest t=1 refers to Jan 1, 1990

What are the values for Y1, Y2, and Y3 in the air series?

Y1 = 35153577 Y2 = 32965187 Y3 = 39993913

Chapter 2 Questions

Plot each of three pre-event time series (Air, Rail, Car).

Travel Prior to the 9/11 Terrorist Attacks

Air.ts<- ts(AirRailCar$AirR, start = c(1990, 1), end = c(2001,8), frequency = 12)
AirPlot<- plot(Air.ts, ylab = "Air Passenger Miles (Millions)", type="l", bty="l", main = "Airline Passenger Miles pre 9/11 Terrorist Attacks") 

Rail.ts<- ts(AirRailCar$RailR, start = c(1990, 1), end = c(2001,8), frequency = 12)
RailPlot<- plot(Rail.ts, ylab = "Rail Passenger Miles (Millions)", type="l", bty="l", main = "Rail Passenger Miles pre 9/11 Terrorist Attacks") 

vehicle.ts<- ts(AirRailCar$VMT, start = c(1990, 1), end = c(2001,8), frequency = 12)
VMTPlot<- plot(vehicle.ts, ylab = "VMT Passenger Miles", type="l", bty="l", main = "Vehicle Passenger Miles pre 9/11 Terrorist Attacks")

What time series components appear from the plot? When it comes to air and rail travel, it is clear that there are multiplicative components. The values vary by percentages. For the vehicle travel, the trend seems to have an additive component, or atleast close.

What type of trend appears? Change the scale of the series, add trend lines, and suppress seasonality to better visualize the trend pattern.

QTRLYAir <- aggregate(Air.ts, nfrequency=4, FUN=sum)
plot(QTRLYAir, xlab = "Year", ylab = "Air Passenger Miles",bty="l", main = "Quarterly Air Miles with Seasonality Suppressed")

QTRLYRail <- aggregate(Rail.ts, nfrequency=4, FUN=sum)
plot(QTRLYRail, xlab = "Year", ylab = "Rail Passenger Miles",bty="l", main = "Quarterly Rail Miles with Seasonality Suppressed")

QTRLYvehicle <- aggregate(vehicle.ts, nfrequency=4, FUN=sum)
plot(QTRLYvehicle, xlab = "Year", ylab = "Vehicle Passenger Miles",bty="l", main = "Quarterly Vehicle Miles with Seasonality Suppressed")

Shipments of Household Appliances

Create a well formatted time plot of the data.

appliancedata <- read.csv("17_0206_Appliance_ShipmentsRev.csv")
appliance.ts <- ts(appliancedata$Shipments, start = c(1985,1), frequency = 4)
appliance.ts <- ts(appliancedata$Shipments, start = 1985, frequency = 4)
applianceZ.ts <- zooreg(appliancedata$Shipments, start = as.yearqtr("1985-1"), frequency = 4)
Window_appliance.ts <- window(applianceZ.ts, start = as.yearqtr("1985-1"), end = as.yearqtr("1987-4"))
plot(applianceZ.ts, col="green", lty=1, lwd=2, xlab = "Year", ylab = "Appliance Shipments ($ Millions)", main = "Quarterly Appliance Shipments")

Which of the four components (level, trend, seasonality, noise) seem to be present in this series?

There seems to be some clear seasonality and potential noise in this series. In the beginning of all of the years, except 1988, there seems to be a slow start to shipments. It’s possible there could be some missing or poor data in 1988 with the consistent trend that has taken place through other years. It would be nice to see how this data was collected or what was going on during that time. By the middle of the year, it seems that the shipments start to hit their stride and peak then slowly fall towards the beginning of the next year.

Forecasting Shampoo Sales

Create a well formatted time plot of the data.

shampoo <- read.csv("17_0206_Shampoo_Sales.csv")
shampoo.ts<- ts(shampoo$Shampoo.Sales, start = c(1995, 1), frequency = 12)
plot(shampoo.ts, col="green", lty=1, lwd=2, xlab = "Time", ylab = "Shampoo Sales",main = "Shampoo Sales 1995-1997")

Which of the four components seem to be present in this series?

There seems to be a clear trend throughout this time series. Over time, shampoo sales slowly started to increase. It’s unclear if there’s much noise considering how the trend does not have much variance throughout the time series.

Do you expect to see seasonality in sales of shampoo? Why?

I wouldn’t expect to see seasonality in sales of shampoo, since it is something that’s needed by everyone throughout the entire year. When it comes to appliance shipments, I could imagine that the beginning of the year would be a slow start, since it is right after the holidays.