BFO 7010 Module 2 Simple Forecasting Models Example

Course	BFO-7010 Business Forecasting
Module	Module 2
Paper type	Module 2 written assignment: benchmark forecasting methods and accuracy measures
Length	About 675 words, 5 pages
Format	APA 7 student paper
School	Thomas Edison State University
Program	MBA in Data Analytics
Updated	October 2026

Free sample paper for BFO 7010 Module 2

The Benchmark to Beat: Four Simple Forecasts of Monthly Propane Deliveries

Student Name

MBA in Data Analytics, Thomas Edison State University

BFO-7010: Business Forecasting

Mentor Name

Month Day, Year

What this page is doingThe title states the purpose of simple methods in the course, setting a benchmark every later model must beat, for the composite series used throughout.

The Benchmark to Beat: Four Simple Forecasts of Monthly Propane Deliveries

The Data

Deliveries peak in January, around 600,000 gallons, and fall to about 210,000 in midsummer, a seasonal swing far larger than the slow upward trend. The training period, January 2019 to December 2023, has 60 months with a mean of 371,800 gallons; December 2023 delivered 578,200. A time plot shows the same winter shape each year, with colder winters, such as 2022, standing out.

Part A: Four Benchmark Forecasts

The drift slope is slightly negative only because December 2023 was a little lower than January 2019; on this series, drift mainly carries the last winter value forward, much as the naive method does.

Method	How the 2024 forecast is made
Mean	Every month equals the 60-month average, 371.8
Naive	Every month equals the last observation, 578.2
Seasonal naive	Each month equals the same month of 2023
Drift	The last value extended along the line from the first to the last observation, a slope of -0.47 a month

Part B: Accuracy on the 2024 Test Set

Mean absolute error and root mean squared error are in thousands of gallons; RMSE weighs large misses more heavily. Percentage error is easy to explain but unstable when actual values are small, which is not a problem here. The mean absolute scaled error divides MAE by the in-sample MAE of the seasonal naive method, 32.3, so values below 1 beat that benchmark on the training data (Hyndman & Koehler, 2006). Seasonal naive is far ahead on every measure because it is the only method that captures the seasonal pattern; the naive and drift methods carry a winter value into summer and miss by over 300,000 gallons in July.

Method	MAE	RMSE	MAPE	MASE
Mean	128.0	145.3	34.8%	3.96
Naive	191.3	228.3	67.7%	5.92
Seasonal naive	31.4	38.3	8.3%	0.97
Drift	189.3	225.7	67.0%	5.86

What this page is doingExplaining why the seasonal naive method wins, rather than only reporting that it does, ties the accuracy table back to the shape of the series.

Part C: Checking the Residuals

The seasonal naive residuals, each month minus the same month a year earlier, average 10.2 with a standard deviation of 41.3. The positive mean shows bias: the method ignores the upward trend, so its forecasts run low by about 10,000 gallons a month on average. The first-order autocorrelation of the residuals is 0.28, so each month's error carries information about the next, which a better model could use. Both findings point to the models in later modules, which add trend and weather.

Part D: Using the Benchmark

For 2024, the seasonal naive forecast totaled 4,689,600 gallons against an actual 4,799,300, a 2.3 percent shortfall consistent with the bias found above. Hyndman and Athanasopoulos (2018) recommend keeping such a benchmark in every comparison. Makridakis et al. (2018) found in a large forecasting competition that simple methods are hard to beat by much, so for Pinelands, any model that cannot beat a MASE of about 1 on the test year should not replace the seasonal naive method in planning, however sophisticated it looks.

The test-set comparison has limits. One year is a short test, and 2024 was close to an average winter; a cold or warm year could change the ranking among the stronger models later in the course. A time series cross-validation, refitting each method on an expanding window and forecasting one year ahead from several origins, would give a sturdier comparison and is the approach the textbook recommends when data allow.

Part E: A Rolling-Origin Check

To see whether one test year is misleading, each method was refitted on an expanding window ending in December 2020, 2021, 2022 and 2023 and used to forecast the following twelve months.

The ranking holds in every year, but the seasonal naive error nearly doubles in 2022, the coldest winter, because copying the previous year cannot anticipate a colder one. That pattern is the strongest argument for adding weather as a predictor, and an average error of about 42,000 gallons a month is the figure later models should be judged against.

Forecast year	Seasonal naive RMSE	Mean RMSE
2021	36.4	148.7
2022	60.3	162.7
2023	32.7	159.2
2024	38.3	145.3
Average	42.0	154.0

References

Hyndman, R. J., & Athanasopoulos, G. (2018). Forecasting: Principles and practice (2nd ed.). OTexts.

Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. International Journal of Forecasting, 22(4), 679-688. https://doi.org/10.1016/j.ijforecast.2006.03.001

Makridakis, S., Spiliotis, E., & Assimakopoulos, V. (2018). The M4 Competition: Results, findings, conclusion and way forward. International Journal of Forecasting, 34(4), 802-808. https://doi.org/10.1016/j.ijforecast.2018.06.001

Reading the BFO 7010 Module 2 assignment instructions

The six BFO-7010 written assignments share 40 percent of the grade, and most are built from exercises in Hyndman and Athanasopoulos, with data supplied through links in the assignment. The syllabus asks students to state answers clearly and explain how they reached them. For the simple methods module, expect to produce mean, naive, seasonal naive and drift forecasts, split data into training and test sets, compute MAE, RMSE, MAPE and MASE, and check residuals for bias and autocorrelation. The data here are a composite monthly series built for this page; your assignment will use the textbook's data sets.

Inside the BFO 7010 Module 2 example

The data section describes the series and its seasonal shape. Part A explains how each benchmark forecast is made, with a note on why drift behaves like naive here. Part B compares accuracy on the held-out year with four measures, explains what each measure emphasizes and why seasonal naive wins. Part C examines the best method's residuals for bias and autocorrelation and links the findings to later modules. Part D uses the benchmark for a planning total, states the standard every later model must meet and recommends time series cross-validation as a sturdier test, with Makridakis et al. (2018) on simple methods in context. Part E repeats the comparison from several forecast origins to show the ranking is not an accident of one year.

Reading the BFO 7010 Module 2 grading rubric

The simple methods assignment earns roughly a sixth of the 40 percent assigned to written work. Mentors check correct forecasts from each method, accurate error measures computed on the test set rather than the training data, an explanation of what each measure shows and residual checks interpreted correctly. Explaining results in plain language, such as why a method misses in summer, is part of what the syllabus means by showing how answers were reached. Tables with clear units help, and naming the next step a residual problem suggests shows forecasting judgment. The rolling-origin check in this sample is the kind of extra step that separates strong answers from adequate ones.

BFO 7010 Module 2 help: mistakes that cost marks

Common mistakes in this module are measuring accuracy on the training data, comparing MAPE across series with very different scales without comment and reporting the best method without asking whether its residuals look like white noise. Hold out the final season, compute every measure on it and explain each in a sentence. Check residual mean and autocorrelation before declaring a winner. Keep units in every table. A free first sample of your own BFO-7010 Module 2 problems is available on request. Write the formula for each measure once, so the reader can see what the numbers mean. When the best method still leaves biased residuals, say what kind of model could fix it.

Write yours, or have the desk draft it

This paper is an original model document written by our desk, not a submitted student paper and not an official Thomas Edison State University document. Read it for the moves, then write your own to the instructions in your classroom. If you want one built to your exact prompt and rubric, the first custom sample is free and arrives in 24 to 48 hours.

More BFO 7010 and MBA in Data Analytics sample papers

BFO 7010 Module 2 questions, answered

Where can I find a free BFO 7010 Module 2 sample paper?

The propane series assignment above fits four benchmark methods, compares them on a held-out year with MAE, RMSE, MAPE and MASE and checks the best method's residuals.

What are the four simple forecasting methods?

The mean method, the naive method, the seasonal naive method and the drift method, which extends the line from the first to the last observation.

What does a MASE below 1 mean?

The forecast's average error is smaller than that of the in-sample seasonal naive (or naive) benchmark used for scaling.

Why hold out a test set?

Because accuracy on data used to fit a model overstates how well it will forecast data it has not seen.

What should good forecast residuals look like?

Uncorrelated with mean near zero; patterns or bias mean information is left in the data.

BFO 7010 Module 2 Written Assignment: Simple Forecasting Methods and Accuracy Example