Adaptive approaches to stochastic programming

McAllister, Patrick H.

doi:10.1007/bf02204808

Cited by 23 publications

(13 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Ž . rather than cumulated McAllister 1991, Mookerjhee and Sopher 1994 In the Cross model, strategies have utilities that are weighted averages of past utilities and Ž . Ž .…”

Section: 2mentioning

confidence: 99%

See 1 more Smart Citation

Experience-weighted Attraction Learning in Normal Form Games

1999

View full text Add to dashboard Cite

Ž . In 'experience-weighted attraction' EWA learning, strategies have attractions that reflect initial predispositions, are updated based on payoff experience, and determine Ž . choice probabilities according to some rule e.g., logit . A key feature is a parameter ␦ that weights the strength of hypothetical reinforcement of strategies that were not chosen according to the payoff they would have yielded, relative to reinforcement of chosen strategies according to received payoffs. The other key features are two discount rates, and , which separately discount previous attractions, and an experience weight. EWA Ž . includes reinforcement learning and weighted fictitious play belief learning as special cases, and hybridizes their key elements. When ␦ s 0 and s 0, cumulative choice reinforcement results. When ␦ s 1 and s , levels of reinforcement of strategies are exactly the same as expected payoffs given weighted fictitious play beliefs. Using three sets of experimental data, parameter estimates of the model were calibrated on part of the data and used to predict a holdout sample. Estimates of ␦ are generally around .50, around .8᎐1, and varies from 0 to . Reinforcement and belief-learning special cases are generally rejected in favor of EWA, though belief models do better in some constant-sum games. EWA is able to combine the best features of previous approaches, allowing attractions to begin and grow flexibly as choice reinforcement does, but reinforcing unchosen strategies substantially as belief-based models implicitly do.

show abstract

“…Ž . rather than cumulated McAllister 1991, Mookerjhee and Sopher 1994 In the Cross model, strategies have utilities that are weighted averages of past utilities and Ž . Ž .…”

Section: 2mentioning

confidence: 99%

“…Ž . rather than cumulated McAllister 1991, Mookerjhee and Sopher 1994 . Bush andMosteller 1995 , Cross 1983 , where the updating may be different for losses and gains.…”

mentioning

confidence: 97%

Experience-weighted Attraction Learning in Normal Form Games

1999

View full text Add to dashboard Cite

show abstract

“…1 For example, Beil (1990, 1991), McAllister (1991), Crawford (1995), Roth and Erev (1995), VanHuyck, Battalio and Rankin (2001), Anderson (2000), Camerer (2003), and Wixted (2004aWixted ( , 2004b. 2 For example, see Zimmerman (1982), Argote, Beckman and Epple (1990), Gruber (1992), Bahk and Gort (1993), Marimon and Sunder (1994), McAfee and McMillan (1996), Nye (1996), Sargent (1999), Benkard (2000), Thompson (2001), Thornton and Thompson (2001), Evans, Honkaphoja and Marimon (2001), Evans and Honkaphoja (2001), Barrios and Strobl (2004), List (2003), Harrison and List (2004).…”

mentioning

confidence: 99%

Learning in the Credit Card Market

Agarwal¹,

Driscoll²,

Gabaix³

et al. 2011

SSRN Journal

View full text Add to dashboard Cite

Agents with more experience make better choices. We measure learning dynamics using a panel with four million monthly credit card statements. We study add-on fees, specifically cash advance, late payment, and overlimit fees. New credit card accounts generate fee payments of $15 per month. Through negative feedback -i.e. paying a fee -consumers learn to avoid triggering future fees. Paying a fee last month reduces the likelihood of paying a fee in the current month by about 40%. Controlling for account fixed effects, monthly fee payments fall by 75% during the first three years of account life. We find that learning is not monotonic. Knowledge effectively depreciates about 10% per month, implying that learning displays a strong recency effect.JEL classification: D1, D4, D8, G2.

show abstract

“…In the context of experimental game theory with human subjects, several authors have used simple models of reinforcement learning to successfully explain and predict behaviour in a wide range of games (McAllister, 1991;Roth & Erev, 1995;Mookherjee & Sopher, 1994;Mookherjee & Sopher, 1997;Chen & Tang, 1998;Erev & Roth, 1998;Erev et al, 1999). In general, the various models of reinforcement learning that have been applied to strategic contexts tend to differ in the following, somewhat interrelated, features:…”

Section: Reinforcement Learning In Strategic Contextsmentioning

confidence: 99%