| Course | GB513 Business Analytics |
|---|---|
| Unit | Unit 10 |
| Paper type | Applied forecast analysis |
| Length | About 1,100 words, 6 pages double-spaced |
| Format | APA 7 student paper |
| School | Purdue University Global |
| Program | MBA |
| Updated | October 2026 |
Free sample paper for GB513 Unit 10
Three Forecasts, One Hiring Plan: A Unit 10 Applied Forecast Analysis for a Heating and Cooling Contractor
[Student Name]
Purdue University Global
GB513: Business Analytics
Unit 10 Applied Forecast Analysis
[Instructor Name]
[Date]
The contractor, its service calls and every figure are composite and illustrative; the methods cited are real.
Executive Summary
Summit Air Services should plan next year's technician staffing on a linear trend adjusted by seasonal indexes, which forecast this year's service calls within about 2 percent when tested on data it had not seen. Next year's forecast is about 1,369 calls in the first quarter, 1,010 in the second, 1,698 in the third and 1,191 in the fourth. With five year-round technicians handling about 210 calls each per quarter, Summit should hire two seasonal technicians for the first quarter, none for the second, three or four for the third and one for the fourth.
The Decision
Summit Air Services repairs and maintains furnaces, heat pumps and air conditioners for homes and small businesses. Demand peaks in summer, when air conditioners fail, and again in winter, when furnaces fail. Summit employs five year-round service technicians and hires seasonal technicians for peaks. The operations manager, Carla Nguyen, must decide how many seasonal technicians to hire for each quarter of next year.
The Data
Service calls for the last twelve quarters, from the oldest, were: year one, 1,180, 860, 1,420 and 990; year two, 1,260, 910, 1,530 and 1,050; year three, 1,330, 975, 1,610 and 1,120. A line chart shows a steady upward trend with peaks in the first and third quarters each year. Rob Hyndman and George Athanasopoulos recommend plotting a series before choosing a method, since trend and seasonality determine which methods can work (Hyndman & Athanasopoulos, 2021).
Testing the Methods
To compare methods fairly, each was used to forecast year three's four quarters, then scored against the calls Summit actually received. Mean absolute deviation measures the average size of the errors in calls, and mean absolute percentage error measures them as a share of actual calls; David Levine and colleagues recommend comparing methods on a holdout period with measures like these rather than on the data used to fit them (Levine et al., 2021).
Four-quarter moving average. Each quarter's forecast is the average of the four quarters before it. Forecasts for year three were about 1,188, 1,205, 1,221 and 1,241. Because a four-quarter average smooths away the seasons, it under-forecast the peaks and over-forecast the troughs. Mean absolute deviation was about 221 calls and mean absolute percentage error about 17.3 percent.
Exponential smoothing, smoothing constant 0.3. Each forecast combines 30 percent of the last actual value with 70 percent of the last forecast. Forecasts for year three were about 1,170, 1,218, 1,145 and 1,285. Simple exponential smoothing has no seasonal component, so it lagged behind every turn in the series. Mean absolute deviation was about 258 calls and mean absolute percentage error about 20.1 percent.
Linear trend with seasonal indexes. Using years one and two, seasonal indexes were computed as each quarter's average divided by the overall average: about 1.061 for the first quarter, 0.770 for the second, 1.283 for the third and 0.887 for the fourth. The data were deseasonalized by dividing by these indexes, a linear trend was fitted, and the trend was multiplied back by the indexes. Forecasts for year three were about 1,287, 944, 1,592 and 1,113. Mean absolute deviation was about 25 calls and mean absolute percentage error about 2.0 percent.
Only the seasonal trend model follows both the rise and the seasons, and its error is about a tenth of the other two methods' errors.
Choosing the Method
The seasonal trend model wins clearly on both measures. James Evans notes that a forecasting method should match the patterns in the data, and that methods without a seasonal component will miss regular peaks (Evans, 2020). Summit's data have both trend and strong seasonality, so a method that captures both is the natural choice.
Next Year's Forecast
The model was refitted on all twelve quarters. The updated seasonal indexes are about 1.059, 0.771, 1.281 and 0.888, and the deseasonalized trend rises about 16 calls a quarter. Next year's forecasts are about 1,369 calls in the first quarter, 1,010 in the second, 1,698 in the third and 1,191 in the fourth.
The Hiring Plan
Each technician handles about 210 calls a quarter. Dividing each forecast by 210 gives the technicians needed: about 6.5 in the first quarter, 4.8 in the second, 8.1 in the third and 5.7 in the fourth. With five year-round technicians, Summit should hire two seasonal technicians for the first quarter, none for the second, three for the third with overtime as a cushion or four if overtime is not available, and one for the fourth.
How Sensitive Is the Hiring Plan?
The plan depends on the forecast, so it was tested against the model's recent error. In the holdout year, the seasonal trend model missed by about 2 percent on average and by at most about 4 percent in any quarter. If next year's third quarter came in 4 percent above forecast, about 1,766 calls, Summit would need about 8.4 technicians, still covered by four seasonal hires. If it came in 4 percent below, about 1,630 calls, three seasonal hires would be enough. The first and fourth quarters are similarly stable: a 4 percent miss would not change the number of seasonal hires in either. The plan is therefore robust to errors of the size the model has made so far, though not to a very unusual summer.
The Cost of Getting It Wrong
Errors in each direction cost money. Hiring one seasonal technician too many for a quarter costs about $14,000 in wages and benefits. Hiring one too few means about 210 calls a quarter either handled on overtime, at about $9,000 in premium pay, or answered late, which risks losing customers to competitors during the peak. Because a late response in a heat wave can lose a customer for years, the plan leans toward the higher figure in the third quarter.
Limitations
Twelve quarters is a short series, and an unusually hot summer or cold winter could push calls well above the forecast. The model assumes the trend continues, which depends on Summit keeping its customer growth. And calls per technician may change if Summit adds new service lines.
Recommendation
Adopt the seasonal trend forecast, hire seasonal technicians as above, recruit for the third quarter by April and rerun the model at the end of every quarter with the newest actual figure added.
References
Evans, J. R. (2020). Business analytics: Methods, models, and decisions (3rd ed.). Pearson.
Hyndman, R. J., & Athanasopoulos, G. (2021). Forecasting: Principles and practice (3rd ed.). OTexts.
Levine, D. M., Stephan, D. F., & Szabat, K. A. (2021). Statistics for managers using Microsoft Excel (9th ed.). Pearson.
What the GB513 Unit 10 assignment asks for
The final GB513 unit asks you to forecast a business series and turn the forecast into a decision. Open with an executive summary stating the method chosen, the forecast and the decision. Describe the decision and the data, and plot the series to show trend and seasonality. Test several methods, such as a moving average, exponential smoothing and a trend with seasonal indexes, on the same holdout period, and compare them on mean absolute deviation and mean absolute percentage error. Choose the method that fits the data's patterns and explain why. Refit it on all the data and produce the forecast. Convert the forecast into a plan, test how sensitive the plan is to forecast error and weigh the cost of errors in each direction. Close with limitations and a recommendation. Expect six to eight pages, citing forecasting and analytics sources in APA 7.
How the GB 513 Unit 10 example is put together
The executive summary recommends the seasonal trend model and gives next year's forecasts and hires. The data section lists Summit Air Services' twelve quarters and cites Hyndman and Athanasopoulos on plotting before choosing a method. Each method forecasts year three: the moving average misses the seasons with an error near 17.3 percent, exponential smoothing lags with about 20.1 percent, and the seasonal trend model, with indexes of about 1.061, 0.770, 1.283 and 0.887, errs by about 2.0 percent, compared on a holdout as Levine, Stephan and Szabat recommend; the highlighted sentence explains the gap. Evans supports matching method to pattern. The refitted model forecasts next year, and 210 calls per technician gives hires. Sections test a 4 percent miss and weigh about $14,000 for an extra hire against overtime and lost customers.
GB513 Unit 10 rubric: what earns full points
GB513 instructors grade this analysis on fair method comparison and a decision grounded in the forecast. An executive summary that states the method, forecast and decision earns early credit. Plotting the series and naming its patterns shows the method choice is informed. Testing every method on the same holdout and comparing error measures carries significant weight. Explaining why the winning method fits the data's trend and seasonality shows understanding. Converting the forecast into a staffing plan, testing its sensitivity to error and weighing the costs of over- and under-staffing show managerial judgment. Limitations and a clear recommendation complete the report. Hyndman and Athanasopoulos's Forecasting: Principles and Practice, Levine, Stephan and Szabat and Evans form the citations in APA 7.
GB513 Unit 10 help: mistakes that cost points
Many GB513 Unit 10 papers choose a forecasting method without comparing it with others on data it has not seen. Others compare methods on the data used to fit them, which flatters complex models. A common gap is the decision: a forecast that never becomes a staffing or inventory plan leaves the analysis unfinished. Some writers ignore how errors would change the plan. Summarize first, plot the series, test methods on a holdout, compare errors, choose and justify a method, refit and forecast, convert to a plan, test sensitivity, weigh error costs and state limits.
Get a GB513 Unit 10 example written to your instructions
Your capstone dataset and GB513 Unit 10 instructions, plus the rubric, are enough for a custom forecast analysis matched to your scenario, whether it predicts sales, demand or staffing. The first sample costs nothing. Its recommendation converts the forecast into the decision your prompt names, and delivery takes 24 to 48 hours. The paper above is an original model document written by our desk, not a submitted student paper and not an official Purdue University Global document.
More GB513 papers and MBA samples
- GB513 Unit 1: Descriptive Statistics Exercise
- GB513 Unit 2: Data Summary Analysis
- GB513 Unit 3: Probability Problem Set
- GB513 Unit 4:
- GB513 Unit 5: Confidence Interval Analysis
- GB513 Unit 6: Hypothesis Test Memo
- GB513 Unit 7: Two-Sample T Test Comparison
- GB513 Unit 8: Simple Regression Analysis
- GB513 Unit 9: Multiple Regression Model
- GB512 Unit 10: Communication Course Reflection
- GB500 Unit 10: Integrative Capstone Paper
GB513 Unit 10 questions, answered
Which forecasting method does the final analysis need to use?
Whichever the prompt names, and if it names none, more than one. The strongest GB513 finals compare at least two methods on the same holdout data and choose on measured error. A moving average, exponential smoothing and a trend with seasonal adjustment cover most business series. Regression-based forecasts also work when the prompt supplies a driver variable.
What is the difference between MAD and MAPE?
MAD is the average size of forecast errors in the original units, such as calls or dollars. MAPE expresses the same errors as a percentage of actual values, which makes it easier to compare across series of different sizes. Reporting both is common. The contractor forecast uses MAPE in the executive summary because an owner grasps percentages faster than raw counts.
How long should the Unit 10 analysis be?
Length follows the prompt, but completed analyses in this course often run six to ten pages plus appendices. What matters more is that the executive summary stands alone, each method gets a short section, and the recommendation is specific. Adding pages of pasted output does not add credit and can hide the decision the rubric is looking for.
Where can I find a free GB513 Unit 10 sample paper?
This page carries a full GB513 Unit 10 applied forecast analysis for an HVAC contractor's service calls, comparing three methods on a holdout year and building a seasonal hiring plan, with APA 7 references.
Why does the moving average forecast so poorly?
A four-quarter moving average smooths away the seasons, so it under-forecasts the summer and winter peaks and over-forecasts the troughs. The contractor's calls swing sharply by season, which the seasonal trend model captures.