ANLY482 AY2017-18T2 Group18/Project Findings

From Analytics Practicum
Revision as of 22:05, 13 April 2018 by Zqlow.2014 (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

TeamDAcctnew.png

Home About Us Project Overview Project Findings Project Management Documentation ANLY482 Homepage


Evaluation of Results & Discussion

The final regression equation for model (i), (ii), (iii) and (iv) are as follows:

Project evluation.png


Reviewing on the overall results of the regression model performed, we can see that:

  • R-squared value for regression model (i) is 0.9676. This shows that with the following independent variables mentioned above, 96.76% of the variation in total project cost could be explained by our model.
  • R-squared value for regression model (ii) is 0.6381. This shows that with the following independent variables mentioned above, 63.81% of the variation in manpower headcount could be explained by our model. Further investigation by LS 2 should be done to find out other variables that might contribute to the variation in manpower headcount as a cost driver, to better estimate manpower costs.
  • R-squared value for regression model (iii) is 0.9614. This shows that with the following independent variables mentioned above, 96.14% of the variation in chemical & materials costs could be explained by our model.
  • R-squared value for regression model (iv) is 0.9147. This shows that with the following independent variables mentioned above, 91.47% of the variation in equipment costs could be explained by our model.

While the models have shown significant predictor factors for total project cost, manpower headcount, chemical & materials costs and equipment costs, we believe that the current iteration of models performed can be strengthened further in terms of its explanatory and predictive ability. From our confirmatory analysis, we have tried to derive other relevant variables to be included in the models. With the low number of observations gathered due to the limited number of conservancy projects LS 2 undertakes, the analysis may benefit by incorporating greater years’ worth of data from the current dataset gathered. Another limitation when performing regression analysis is that the regression equation is only valid from the range of the independent variables identified. Trying to estimate the dependent variable outside of the range of the independent variables will be an extrapolation, and possibly wrong.



Recommendation & Future Directions for our Project Sponsor
  • We recommend our project sponsor to use the model as a complementary tool of cost-estimation when determining the bid amount to place. This model should be used in conjunction with the management’s past knowledge and experience as the model is not a predictive but an explanatory model. In addition, management should take note of the limitation of the model and do not rely on the model in totality. Also we suggest similar analysis to be done for MOE projects over time with more accumulated data and for Housekeeping projects by identifying common factors that are available in the tender document across various types of tender document formats and information disclosed.
  • To further strengthen the explanatory power of the model, it is imperative that more information are fed to the model. Hence, we suggest strengthening the ERP system that LS 2 currently uses for completeness of data LS 2 collects, storage of data in more structured manner and easy retrieval of data. With that, LS 2 also needs to advocate stricter use and storage of data on the ERP system to its staff. This more integrated database is not only good for data collection but also useful for the management to monitor its costs and various projects.
  • The bid amount is decided with great effort and consideration and it is directly linked to the profitability of the company. Hence the bid amount can be used as a budget for each project. However, such budget monitoring system is not utilised by the management and the management does not actively monitor the cost real time as the project progresses. The management should compare the bid placed with cost incurred to date to control expenses and hence to minimise the chance of making lower-than-expected profit or even loss. This is possibly enabled by using a dashboard which can provide a high-level overview of the projects and their ‘profit and loss’ as well as ‘budget versus actual expenses’ implications.