Cost-effectiveness

GuilleGuille Member
Hello all,

I have a doubt related to RIOS Cost-effectiveness.

I am modelling with different activity costs (from 1,200 up to 6500) and using both, the floating budget option and also allocating the budget to the different activities. 

I wanted to understand a bit more about the importance of activity costs in RIOS, more specifically about its preference for the lowest cost-activities (when using the floating budget option). Is there any news related to this topic?

Thanks a lot!

Guille.

Comments

  • Hi Guille -- 

    The behavior you have discovered is due to a structural issue with RIOS, where values are normalized with respect to various weights at every step of the calculations, except the last step, where normalized values between zero and one are all divided by a real cost vector to generate the cost-effectiveness score -- which is then used to rank interventions for selection in the portfolio. This effectively means that, in the absence of the user implementing work-arounds (or possibly extremely skewed performance), RIOS will always pick the cheapest thing, at least to start. 

    In this respect, the language surrounding RIOS is often a bit overstated and we have been trying to walk it back here and there. In general we consider the RIOS intermediate outputs as much more valid for assessing the relative effectiveness in space of each individual intervention (ie, where on the landscape a given intervention makes more sense or less sense), rather than comparing the cost-effectiveness across interventions. 

    In the meantime, there are a few options you might consider depending on your needs:
    1) The most recent release of RIOS (1.1.16) includes a "fractional activity effectiveness" so that different activities can have different levels of effectiveness in bringing about a particular transition. I'm not sure exactly how this translates within the model structure, but I think it should help with the problem you face, unless there's still a lot of normalization after the point at which the fractional activity weighting is implemented. 

    2) Depending on your application (in particular, what objectives and how many interventions you have), you might consider utilizing our alternate prioritization tool ROOT, which is in many was an advancement on RIOS in terms of the optimization algorithm, but is less tailor-made for watershed targeting with multiple interventions -- in particular it requires the user to provide their own estimates of "impact potential" that each intervention could have on each objective (roughly analogous to the activity scores in RIOS). The user-interface version also is currently built to work with a single intervention at a time, though custom scripting can get around that. 

    3) Some of us are currently building up a script-based workflow to achieve most of the functionality of RIOS combined with the improved optimization of ROOT. It will probably be quite some time before this works its way into ROOT or another NatCap product, but I would be happy to talk more about it if you have near terms needs and we can see what we can do. 

    Best,
    -Ben




  • Thanks a lot for your reply Ben! I am working with RIOS latest version and doing what you propose on the first option.I will keep you posted because the other options you propose are definitely interesting. Additionally, I was in contact with Stacie and she also helped me to implement a scenario without costs (by giving the activities a cost of 1).

    Guille.
  • Hi Ben,

    Can I ask a bit more about this fractional effectiveness of activities - is this specified in the Objective-Transition Weights?
    If so, how can you specify this for multiple actions listed under the same Transition?

    Secondly, do you know how the different types of transitions are treated in the calculations. For example, I have two widths of buffer (5,25m) that I would like RIOS to prioritize - and they are allowed on the same pixel, but have different efficiencies with respect to Erosion control and Nutrients (which I specify in the Obj-Transition weights). I've tried running the model twice, with one action listed as Reveg (assisted) and the other as Ag veg Mgt without any costs assigned to see where each is most effective; and in the second run I flip which transition type each activity was assigned to. I get two different results. The activity listed under Agricultural Vegetation Management is always selected more often. Why is this?

    Could someone explain or point me towards the part of the user guide which details how transitions are treated in the ranking and prioritization?

    Thanks,
    Sylvie   

  • swolnyswolny Member, NatCap Staff
    Hi Sylvie -

    The "fractional effectiveness of activities" is specified in the Transition Potential tab. Apologies that the explanatory text in there hasn't been updated, but you can enter decimal values between 0 and 1 here to indicate that different activities are more or less effective at causing a transition.

    As for how transitions are treated. You can look at Figure 6 for the big picture of what feeds into what. Page 29 is where equations are listed showing how the transition weights affect objective scores etc. I'm not sure if it provides the detail you're looking for though. 

    It isn't immediately obvious to me why the Ag Vegetation Management activity would always be selected, really, it should be treated the same numerically as Revegetation, unless you've specified transition weights that would favor one over the other. You say that you're specifying obj-transition weights - are you also flipping those when you flip the definition of the buffers between Reveg and Ag Management?

    ~ Stacie
  • RichRich Administrator, NatCap Staff
    It's possible if they are exactly the same weights and costs you're getting Ag Vegetation Management because it's alphabetically first?
Sign In or Register to comment.