Proof of Theorem 1, p. ec2 — the weights (EC.5), λ_t = ∏_j P(o^t_j | S_j), are nonnegative and sum to one
ProvedDecisionForest.Universal.prodWeight_isDistributiondiscrete-choicep2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1probability-distribution
Let be a discrete choice model on products. For a tuple of purchase options, one per assortment, define as in (EC.5)
Then for every tuple , and
the sum running over all tuples.
These are the weights the proof of Theorem 1 puts on the trees ; the statement says they form a probability distribution.
Formalization Note Tuples are functions o : Finset (Fin N) → Option (Fin N) and is prodWeight P o.
Differences from the source:
- encoding: trees indexed by through an enumeration of tuples → weights indexed by the tuples themselves; reason: same family of numbers.
- encoding: products →
Fin N, options →Option (Fin N)withnonethe option ; reason: Lean convention, same instances.
Preamble
import Mathlib import Definitions.Def_DecisionForest_Universal_Setting
Formal statement
namespace DecisionForest.Universal
theorem prodWeight_isDistribution (N : ℕ) (P : Option (Fin N) → Finset (Fin N) → ℝ)
(hP : IsChoiceModel P) :
(∀ o : Finset (Fin N) → Option (Fin N), 0 ≤ prodWeight P o) ∧
∑ o : Finset (Fin N) → Option (Fin N), prodWeight P o = 1 := by sorry
end DecisionForest.Universal
Source
Chen and Mišić, Decision Forest: A Nonparametric Approach to Modeling Irrational Choice, arXiv:1904.11532v3, p. ec2 (PDF 31), proof of Theorem 1, (EC.5)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.