We are pleased to introduce IBM Watson Tradeoff Analytics, a new cognitive service that takes the guesswork out of decision making, available to the development community through the Watson Developer Cloud.
The first role of cognitive computing is to really enable humans to make complex decisions. This is one of humans’ most important cognitive tasks, and we do it all the time. If we look up the definition of “Decision Making” in Wikipedia, the first sentence tells us that it has to do with “ … selection of a course of action among several alternatives”. Why is decision making so difficult? Can’t we just select the best alternative? Where does the dilemma come from? The answer resides in the inherent nature of hard choices. In hard choices, there is no correct or best answer. One alternative is better in one way, the other alternative is better in a different way, and neither is better than the other overall. For example, if we are considering an investment in mutual funds, we want to compare which is best across the following performance time dimensions, as well as their risk:
Year To Date Performance, Short Term (1 year) Performance, Mid Term (3 years) Performance, Long Term (10 years) Performance, and Risk.
Let’s say there are 115 mutual funds to start with. Using the prevalent tools on investment sites will always leave you overwhelmed. The lists are long, with few filters, and no real way to examine your tradeoffs. The probability is that you will unintentionally ignore precious investible assets!
Let’s see how IBM Watson’s Tradeoff Analytics can be used to approach this task:
1) Question Editor:
This part of the tool reflects your decision question, which in our case is to choose the best fund that meets your criteria.
You can further work with these goals so that they reflect your true preferences. For example, you can choose to delete the “Short term” criterion, and work with the remaining 4.
Potentially you could press the “Add” link to add other objectives that do not appear here and that you care about.
Tradeoff Analytics reduces the problem of 115 funds to 15 choices that were found optimal, based on your criteria. Each fund is shown as a radial chart. The amount of color in each slice is different, so the fuller the slice the better the fund is on the associated Objective.
Each objective is represented by a vertex in the polygon. The funds are “pulled” by the objectives according to their values. Closer to the vertex is better with respect to that objective.
3) Auto Excluded funds:
Funds that are inferior in all the objectives were auto-excluded by Tradeoff Analytics. Here you can understand why OSVLX does not appear on the Tradeoff Analytics map. Each fund is shown as a zigzag line, and each objective is represented by a vertical bar.
The OSVLX fund is represented by the gray line, and the funds OSAIX, OPHAX, and OSPHX exceed OSVLX across all objectives.
4) The 5 filters on the left hand side correspond to the 5 objectives that we want our fund to meet. For each objective we have a filter, and you can use the filters to put a threshold on some of the objectives, thereby scaling your selection to fewer options. Again, we are filtering among the optimal options that have tradeoffs between them.
5)Tradeoff Analytics makes automatic recommendations for alternatives where small sacrifices in one criteria lead to large gains in others. You can now select your favorite fund and examine it in further detail. Tradeoff Analytics will help with a “Did you know?” type of insight.
6)Tradeoff Analytics helps you minimize regret by highlighting options you might have missed. This sensitivity analysis tool shows you alternatives that may be more appealing than the one you selected. For example, once you selected OSCSX, you are advised that OSAIX is also optimal, but with small compromise on Long Term production (less than 1%), you will get relatively large positive gains in all the rest of your objectives: Nearly 40% more in YTD, about 18% more in Short Term, 5% in Mid Term, and less Risk.
Tradeoff Analytics takes the guesswork out of complex decision-making so you can make educated choices. You are more likely to be satisfied with your decision process and feel more confident in going ahead with the choice made. With time Tradeoff Analytics will leverage the users’ decision paths to derive insights that will refine and improve its decision assistance support, making it more personal and domain specific.