Purpose: Learn who to import data in R. Employ the forecast package in R to obtain simple forecasts. In particular, apply the methods of mean, naïve, and seasonally naïve forecasts and, furthermore, assess their out-of-sample accuracy using MAE, RMSE, and MAPE. Deadline: Tuesday, September 19, 2017

Introduction

It is the beginning of January 2018. ECO5375 Inc. had the following sales in $US from Jan 2016 to December 2017

Year Month Sales Year Month Sales Year Month Sales
2016 Jan 1,000 2017 Jan 2,000 2018 Jan ???
2016 Feb 2,000 2017 Feb 3,000 2018 Feb ???
2016 Mar 3,000 2017 Mar 4,000 2018 Mar ???
2016 Apr 4,000 2017 Apr 5,000 2018 Apr ???
2016 May 5,000 2017 May 6,000 2018 May ???
2016 Jun 6,000 2017 Jun 7,000 2018 Jun ???
2016 Jul 7,000 2017 Jul 8,000 2018 Jul ???
2016 Aug 8,000 2017 Aug 9,000 2018 Aug ???
2016 Sep 9,000 2017 Sep 10,000 2018 Sep ???
2016 Oct 10,000 2017 Oct 11,000 2018 Oct ???
2016 Nov 11,000 2017 Nov 12,000 2018 Nov ???
2016 Dec 12,000 2017 Dec 13,000 2018 Dec ???

The data is available as a .csv-file here

Your line manager at ECO5375, Inc. Mr. Kris Hock wants you to forecast 2018 sales month by month using three different forecast methods.

Question 1

Forecast Jan 2018 to Dec 2018 using the mean forecast method. (1.0 point)

Question 2

Forecast Jan 2018 to Dec 2018 using the naïve forecast method. (1.0 point)

Question 3

Forecast Jan 2018 to Dec 2018 using the seasonally naïve forecast method. (1.0 point)

Jumping forward to you annual bonus discussion in January 2019, with the sales data from January 2018 to December 2018 finally recorded. From January 2018 to December 2018 sales turn out to be:

Year Month Sales Year Month Sales Year Month Sales
2016 Jan 1,000 2017 Jan 2,000 2018 Jan 3,000
2016 Feb 2,000 2017 Feb 3,000 2018 Feb 4,000
2016 Mar 3,000 2017 Mar 4,000 2018 Mar 5,000
2016 Apr 4,000 2017 Apr 5,000 2018 Apr 6,000
2016 May 5,000 2017 May 6,000 2018 May 7,000
2016 Jun 6,000 2017 Jun 7,000 2018 Jun 8,000
2016 Jul 7,000 2017 Jul 8,000 2018 Jul 9,000
2016 Aug 8,000 2017 Aug 9,000 2018 Aug 10,000
2016 Sep 9,000 2017 Sep 10,000 2018 Sep 11,000
2016 Oct 10,000 2017 Oct 11,000 2018 Oct 12,000
2016 Nov 11,000 2017 Nov 12,000 2018 Nov 13,000
2016 Dec 12,000 2017 Dec 13,000 2018 Dec 14,000

Question 4

Calculate the MAE, RMSE, and MAPE for your mean forecast. (1.0 point)

Question 5

Calculate the MAE, RMSE, and MAPE for your naïve forecast. (1.0 point)

Question 6

Calculate the MAE, RMSE, and MAPE for your seasonally naïve forecast. (1.0 point)

February 2018: It turns out there was an accounting software error and actual sales were different from what we thought. From January 2018 to December 2018 actual sales turn out to be:

Year Month Sales Year Month Sales Year Month Sales
2016 Jan 1,000 2017 Jan 2,000 2018 Jan 7,000
2016 Feb 2,000 2017 Feb 3,000 2018 Feb 7,000
2016 Mar 3,000 2017 Mar 4,000 2018 Mar 7,000
2016 Apr 4,000 2017 Apr 5,000 2018 Apr 7,000
2016 May 5,000 2017 May 6,000 2018 May 7,000
2016 Jun 6,000 2017 Jun 7,000 2018 Jun 7,000
2016 Jul 7,000 2017 Jul 8,000 2018 Jul 7,000
2016 Aug 8,000 2017 Aug 9,000 2018 Aug 7,000
2016 Sep 9,000 2017 Sep 10,000 2018 Sep 7,000
2016 Oct 10,000 2017 Oct 11,000 2018 Oct 7,000
2016 Nov 11,000 2017 Nov 12,000 2018 Nov 7,000
2016 Dec 12,000 2017 Dec 13,000 2018 Dec 7,000

Question 7

Re-calculate the MAE, RMSE, and MAPE for your mean forecast. (1.0 point)

Question 8

Re-calculate the MAE, RMSE, and MAPE for your naïve forecast. (1.0 point)

Question 9

Re-calculate the MAE, RMSE, and MAPE for your seasonally naïve forecast. (1.0 point)

Question 10

What forecast is the best one on the revised sales data from February 2019. (1.0 point)