nemsys llc - multiple regression
TRANSCRIPT
A regression analysis by:Christopher Pappas
Gregory DavisMalcolm Campbell
Iris HuAmanda Zabriski
Predict the monthly engineer hours required to service a prospective client
Better objectify certain cost factors Utilize results to assist NEMSYS in
increasing efficiency and/or effectiveness
Every business today needs computer technology
Impractical for every company to hire the proper employees needed to maintain working technology
Service companies such as NEMSYS provide a cost-effective and efficient way to keep technology in working order
Interviewed executives at NEMSYS to understand the main drivers of engineer hours
Collected NEMSYS client data Breakdown of monthly service hours for past 2
years Collected predictor data Performed regression analysis
The regression equation is: AMH = 27.0 - 14.1 S + 0.492 WS + 0.69 NP + 5.53 AS - 13.0 NC + 0.201 NP2
AMH = avg monthly engineer hours S = # of servers WS = # of workstations NP = # of network printer AS = avg savvy NC = avg network complexity NP2 = network printer squared
Lawfirm Average age of workstations Ratio of laptops to overall workstations
1050-5-10
99
90
50
10
1
Residual
Perc
ent
5040302010
10
5
0
-5
-10
Fitted Value
Resi
dual
1050-5
4
3
2
1
0
Residual
Fre
quency
151413121110987654321
10
5
0
-5
-10
Observation Order
Resi
dual
Normal Probability Plot Versus Fits
Histogram Versus Order
Residual Plots for average month hrs
Analysis:Predictor Coef SE Coef T PConstant 26.96 13.25 2.04 0.076S -14.092 6.361 -2.22 0.058WS 0.4918 0.1158 4.25 0.003NP 0.687 3.276 0.21 0.839AS 5.527 4.353 1.27 0.240NC -13.041 6.586 -1.98 0.083NP^2 0.2012 0.4468 0.45 0.664
S = 6.35500 R-Sq = 81.5% R-Sq(adj) = 67.6%
Analysis of Variance
Source DF SS MS F PRegression 6 1423.56 237.26 5.87 0.013Residual Error 8 323.09 40.39Total 14 1746.65
Limited in the amount of data available Based on the rule of 6, the minimal
amount of data to be used in the model should be 84 clients NEMSYS is a small company; does not service
that many clients monthly Fewer observations skews the R-squared
towards 1, but you really haven’t explained the variation
Predict the monthly engineer hours required to service a prospective client AMH = 27.0 - 14.1 (1) + 0.492 (20) + 0.69 (2) + 5.53
(1) - 13.0 (0) + 0.201 (22) = 30.45 * $85/hour = $2,588.59
Prediction interval: (16.59, 43.43) * $85/hour = ($1,410.15, $3,691.55)
Conclusion: more data needed Better objectify certain cost factors
YES Utilize results to assist NEMSYS in
increasing efficiency and/or effectiveness YES
Used a squared predictor
Get more data