I calculate the efficiency of public transportation vehicles via the DEA method and I use DEA frontier. It calculates the efficiency of all DMUs as 1. Can anyone identify what the problem is and know how I can fix it?
This is likely to be because you had too few decision making units for the number of inputs and outputs in you DEA model, rather than that they are all actually 100% efficient. Unfortunately as you do not say how many units you have or what your model for their efficiency is it is difficult to tell.
This problem can be addressed in two main ways
1. Increase the number of DMUs
In your case this could be including more vehicles in your analysis, or if you are already including all comparable vehicles changing the time frame so perhaps you consider the weekly use rather than monthly to get 4 times as many observations in each month
2 Decrease the number of inputs and outputs
This is sometimes done using aggregate variables, so if there say maintenance cost, running cost and lease cost for each vehicle you might replace this by total costs
The paper pitfalls and protocols in DEA would be a good read to get started. Dyson, R.G., Allen, R., Camanho, A.S., Podinovski, V.V., Sarrico, C.S. and Shale, E.A., 2001. Pitfalls and protocols in DEA. European Journal of operational research, 132(2), pp.245-259.
Based on the rule of thumb in DEA methodology, the number of DMUS (n) must be considered as follows:
n= > max{m*s, 3(m+s)}
where m and s stand for input and output items. So, if your dataset cannot satify this rule, then you should apply the full ranking DEA models to get more realistic results!
- Badges
- Science topic