My organization has recently got hold of some data which gives propensity-modelled scores per person for various continuos attributes. However, these attributes are all grouped ranges, with the propensity scores for each adding up to one for each person in the data. For instance:

personid assets_0 assets_0_to_1000 assets_1000_to_10000 assets_10000+
1        0.6      0.25             0.13                 0.02
2        0.3      0.35             0.34                 0.01
3        0.08     0.12             0.25                 0.55

Is there any way to use these scores to estimate the value of a persons assets as a continuous variable?

So far I've tried using a weighted formula approach (0*p1 + 1000*p2 + 10000*p3 + 20000*p4), but I wasn't happy with this from a mathematical standpoint.

I've also tried plotting a "density plot" for each person using the upper bound of the asset against its propensity, and then finding the x-intersect of the maximum value on this graph (which seems a bit dodgy as an approach but gets a continuous number per person).

Any suggestions?


closed as unclear what you're asking by kjetil b halvorsen, Michael Chernick, Peter Flom Jul 13 '18 at 13:40

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.


I don't think there is a way to do this.

You're saying "Person 1 has a 2% chance of having more than \$10,000 in assets." Presumably there is a chance that they have more than $1,000,000 in assets.

You don't know the distribution (or the probabilities) within asset groups, so you don't have enough information to construct a continuous variable. If you were willing to make an assumption about those distributions, you could do it. But that's a heck of an assumption.

  • $\begingroup$ Yes; go back to the raw data an ditch propensity. $\endgroup$ – Frank Harrell Jul 13 '18 at 12:13

Not the answer you're looking for? Browse other questions tagged or ask your own question.