Proceedings of the Third Workshop on Experimental Economics and Machine Learning (EEML 2016), Moscow, Russia, July 18, 2016
This volume contains the papers presented at the Third International Workshop on Experimental Economics and Machine Learning held on July 18, 2016 at the National Research University Higher School of Economics, Moscow. This proceedings concentrates on an interdisciplinary approach to modelling human behavior incorporating data mining and expert knowledge from behav- ioral sciences. Data analysis results extracted from clean data of laboratory ex- periments are of advantage if compared with noisy industrial datasets from the web and other sources. In their turn, insights from behavioral sciences help data scientists. Behavior scientists see new inspirations to research from industrial data science. Market leaders in Big Data, as Microsoft, Facebook, and Google, have already realized the importance of Experimental Economics know-how for their business. In Experimental Economics, although fi nancial rewards restrict subjects pref- erences in experiments, the exclusive application of analytical game theory is not enough to explain the collected data. It calls for the development and evalua- tion of more sophisticated models. The more data is used for evaluation, the more statistical signi fi cance can be achieved. Since large amounts of behavioral data are required to scan for regularities, Machine Learning is the tool of choice for research in Experimental Economics. In some works, automated agents are needed to simulate and intervene in human interactions. This proceeding aims to create a forum, where researchers from both Data Analysis and Economics are brought together in order to achieve mutually-bene fi cial results. This year the workshop has hosted nine regular papers and two research proposals on a variety of topics related to di ff erent aspects of human behavior in games, demography, economy crises, stock markets, etc. Each paper has been reviewed by two PC members at least; all these papers rely on di ff erent data analysis techniques and the presented results are supported by data. The representatives of R&D department of Imhonet company, Vladimir Bo- brikov and Elena Nenova, have presented a keynote talk concerning how to consistently value recommendations produced by recommender systems. We would like to thank all the authors of submitted papers and the Pro- gram Committee members for their commitment. We are grateful to our invited speaker and our sponsors: National Research University Higher School of Eco- nomics (Moscow, Russia), Russian Foundation for Basic Research, and ExactPro. Finally, we would like to acknowledge the EasyChair system which helped us to manage the reviewing process.
This paper presents statistics of a controlled laboratory gift-exchange-game experiment. These numbers can be used for assumptions about human behavior in analysis of noisy web data. The experiment was described in ‘The Impact of Social Comparisons on Reciprocity’ by Gachter et al. 2012. As already shown in relevant literature from experimental economics, human decisions deviate from rational payoff maximization. The average gift rate was 31%. Gift rate was under no conditions zero. Further, we derive some additional findings and calculate their significance.
We propose a general equilibrium model to study the spatial inequality of consumers and firms within a city. Our mechanics rely on Dixit and Stiglitz monopolistic competition framework. The firms and consumers are continuously distributed across a two-dimensional space, there are iceberg-Type costs both for goods shipping and workers commuting (hence firms have variable marginal costs based on their location). Our main interest is in the equilibrium spatial distribution of wealth. We construct a model that is both tractable and general enough to stand the test of real city empirics. We provide some theoretical statements, but mostly the results of numerical simulations with the real Moscow data.
In this paper, we present a cohort-based classification approach to the churn prediction for social on-line games. The original metric is proposed and tested on real data showing a good increase in revenue by churn preventing. The core of the approach contains such components as tree-based ensemble classifiers and threshold optimization by decision boundary.
The aim of this work is to compare shifts in the consumer behaviour of Russian households since the mid-nineties till nowadays. The research considers the consumer behaviour of the Russians over almost the maximum possible available data RLMS period, focusing on the crisis years. Special attention is paid to analysis of the effects of crises in 1998 and 2008. To reveal effects as shifts in consumer behaviour in the aftermath of two crises panel data analysis is used to estimate QAIDS model. Due to the complete sample attrition observed in RLMS dataset since 1994, pseudo-panel approach is used.
We propose a new quality metric for recommender systems. The main feature of our approach is the fact, that we take into account the set of requirements, which are important for business application of a recommender. Thus, we construct a general criterion, named “audience satisfaction”, which thoroughly describe the result of interaction between users and recommendation service. During the criterion construction we had to deal with a number of common recommenders’ problems: a) Most of users rate only a random part of the objects they consume and a part of the objects that were recommended to them; b) Attention of users is distributed very unevenly over the list of recommendations and it requires a special behavioral model; c) The value of the user’s rate measures the level of his/her satisfaction, hence these values should be naturally incorporated in the criterion intrinsically; d) Different elements may often dramatically differ from each other by popularity (long tail – short head problem) and this effect prevents accurate measuring of user’s satisfaction. The final metric takes into account all these issues, leaving opportunity to adjust the metric performance based on proper behavioral models and parameters of short head problem treatment.
We propose a general equilibrium model to study the spatial inequality of consumers and firms within a city. Our mechanics rely on Dixit and Stiglitz monopolistic competition framework. The firms and consumers are continuously distributed across a two-dimensional space, there are iceberg-type costs both for goods shipping and workers commuting (hence firms have variable marginal costs based on their location). Our main interest is in the equilibrium spatial distribution of wealth. We construct a model that is both tractable and general enough to stand the test of real city empirics. We provide some theoretical statements, but mostly the results of numerical simulations with the real Moscow data.
This study focuses on changing family formation trajectories in the Russian Federation. In European countries, pathways to family ceased being stable several decades ago, while in Russia – as in any post-socialist country – such features of life course deinstitutionalization as postponement of marriage, rising cohabitation, and reordering of events were revealed only in the 1990s and explained from the perspective of the Second Demographic Transition (SDT). Our aim is to demonstrate how family formation trajectories of men and women from different generations were transforming with the incorporation of data mining. The three-wave panel data of the Russian part of the “Generations and Gender Survey” (2004, 2007, 2011; N=5321) and the retrospective data of the survey “Person, Family, Society” (2013; N=4477) are used for achieving this aim. Sequence Analysis shows that generations born after 1970 started to exhibit de-standardized family formation trajectories. As the proportion of Russians who raise children in cohabitation or while single rises, such models of behavior become more widely accepted and practiced in contemporary Russia. Women experience more events in the family trajectory, take steps toward family formation earlier, and stay alone with children more often than men. Matrimonial and reproductive behavior has become diverse, proving that Russia fully exhibits the SDT.
We analyze the possibility of improving the prediction of stock market indicators by adding information about public mood ex- pressed in Twitter posts. To estimate public mood, we analysed frequencies of 175 emotional markers - words, emoticons, acronyms and abbreviations - in more than two billion tweets collected via Twitter API over a period from 13.02.2013 to 22.04.2015. We explored the Granger causality relations between stock market returns of S&P500, DJIA, Apple, Google, Facebook, Pzer and Exxon Mobil and emotional markers frequencies. We found that 17 emotional markers out of 175 are Granger causes of changes in returns without reverse eect. These frequencies were tested by Bayes Information Criteria to determine whether they provide additional information to the baseline ARMAX-GARCH model. We found Twitter data can provide additional information and managed to improve prediction as compared to a model based solely on emotional markers.
We present two examples of how human-like behavior can be implemented in a model of computer player to improve its characteristics and decision-making patterns in video game. At first, we describe a reinforcement learning model, which helps to choose the best weapon depending on reward values obtained from shooting combat situations. Secondly, we consider an obstacle avoiding path planning adapted to the tactical visibility measure. We describe an implementation of a smoothing path model, which allows the use of penalties (negative rewards) for walking through ``bad'' tactical positions. We also study algorithms of path finding such as improved I-ARA* search algorithm for dynamic graph by copying human discrete decision-making model of reconsidering goals similar to Page-Rank algorithm. All the approaches demonstrate how human behavior can be modeled in applications with significant perception of intellectual agent actions.
This research investigates how variation in sociality, or the degree to which one feels belonging to a group, affects the propensity for participation in collective action. By bringing together rich models of social behavior from social psychology with decision modeling techniques from economics, these mechanisms can ultimately foster cooperation in human societies. While variation in the level of sociality surely exists across groups, little is known about whether and how it changes behavior in the context of various economic games. Specifically, we found some socialization task makes minimal group members behavior resemble that of an established group. Consistent with social identity theory, we discovered that inducing this type of minimal sociality among participants who were previously unfamiliar with each other increased social identity, and sustained cooperation rates in the newly formed groups to the point that they were comparable to those in the already established groups. Our results demonstrate that there are relatively simple ways for individuals in a group to agree about appropriate social behavior, delineate new shared norms and identities.