Simple Decision Rule Research Articles

Don't Stop Now: The Sunk Cost Effect in an Incentivized Lab Experiment

Though widely accepted in psychology, the Sunk Cost Effect (SCE) finds little support from recent evidence in economics. The SCE is when individuals continue along an unprofitable course because of resources expended. The present study employs an abstract laboratory setting with a simple optimal decision rule and repeated decisions to estimate the SCE. The unique design, modeled after penny auctions, allows subjects to endogenously accumulate sunk costs, which psychologists claim is an important component. I find evidence that sunk costs increase one’s willingness to continue along an unprofitable course of action by 21.9%. Subjects do not learn to ignore sunk costs. Reference-based theories provide an alternative mechanism for SCE behavior. Though I find evidence against Realization Utility, reference points explain observed behavior better than sunk costs do. This draws into question whether there is a need for a separate SCE or if the Effect is only a special case of Prospect Theory. Proposed potential mediators are responsibility for sunk costs, gain/loss domain of the choice, profitability of the action, and time between sinking costs and taking action.

When to Own Stocks and When to Own Gold

We show that dynamic investment portfolio asset allocation based on secular market cycles outperforms a buy-and-hold portfolio of equities and outperforms a buy-and-hold portfolio of gold over long periods. An objective definition of secular market enables identification of an appropriate ex-ante risk-on or risk-off posture for a portfolio. We construct an objective measure which we term a “secular market indicator (SMI)” using a modified Shiller Cyclically Adjusted Price-Earnings (CAPE) ratio with gold as a reference point. This SMI has slightly greater predictive power than Shiller’s CAPE Ratio in that it provides a consistent threshold signal for secular macroeconomic reversals. Finally, we use the SMI to create a simple decision rule to shift asset allocation between equity and gold depending on the secular market cycle. The resulting portfolio outperforms an all-equity portfolio and an all-gold portfolio over holding periods of 10+ years about 70% of the time, and produces superior risk-adjusted performance about 80% of the time.

Automated Feedback Selection for Robot-Assisted Training

Abstract Robot-assisted training can be enhanced by using augmented feedback to support trainees during learning. Efficacy of augmented feedback is assumed to be dependent on the trainee's skill level and task characteristics. Thus, selecting the most efficient augmented feedback for individual subjects over the course of training is challenging. We present a general concept to automate feedback selection based on predicted performance improvement. As proof of concept, we applied our concept to trunkarm rowing. Using existing data, the assumption that improvement is skill level dependent was verified and a predictive linear mixed model was obtained. We used this model to automatically select feedback for new trainees. The observed improvements were used to adapt the prediction model to the individual subject. The prediction model did not over-fit and generalized to new subjects with this adaptation. Mainly, feedback was selected that showed the highest baseline to retention learning in previous studies. By this replication of our former best results we demonstrate that a simple decision rule based on improvement prediction has the potential to reasonably select feedback, or to provide a comprehensible suggestion to a human supervisor. To our knowledge, this is the first time an automated feedback selection has been realized in motor learning.

Computational framework for risk-based planning of inspections, maintenance and condition monitoring using discrete Bayesian networks

This paper presents a computational framework for risk-based planning of inspections and repairs for deteriorating components. Two distinct types of decision rules are used to model decisions: simple decision rules that depend on constants or observed variables (e.g. inspection outcome), and advanced decision rules that depend on variables found using Bayesian updating (e.g. probability of failure). Two decision models are developed, both relying on dynamic Bayesian networks (dBNs) for deterioration modelling. For simple decision rules, dBNs are used directly for exact assessment of total expected life-cycle costs. For advanced decision rules, simulations are performed to estimate the expected costs, and dBNs are used within the simulations for decision-making. Information from inspections and condition monitoring are included if available. An example in the paper demonstrates the framework and the implemented strategies and decision rules, including various types of condition-based maintenance. The strategies using advanced decision rules lead to reduced costs compared to the simple decision rules when condition monitoring is applied, and the value of condition monitoring is estimated by comparing the lowest costs obtained with and without condition monitoring.

A Physics-Inspired Mechanistic Model of Migratory Movement Patterns in Birds

In this paper, we introduce a mechanistic model of migratory movement patterns in birds, inspired by ideas and methods from physics. Previous studies have shed light on the factors influencing bird migration but have mainly relied on statistical correlative analysis of tracking data. Our novel method offers a bottom up explanation of population-level migratory movement patterns. It differs from previous mechanistic models of animal migration and enables predictions of pathways and destinations from a given starting location. We define an environmental potential landscape from environmental data and simulate bird movement within this landscape based on simple decision rules drawn from statistical mechanics. We explore the capacity of the model by qualitatively comparing simulation results to the non-breeding migration patterns of a seabird species, the Black-browed Albatross (Thalassarche melanophris). This minimal, two-parameter model was able to capture remarkably well the previously documented migration patterns of the Black-browed Albatross, with the best combination of parameter values conserved across multiple geographically separate populations. Our physics-inspired mechanistic model could be applied to other bird and highly-mobile species, improving our understanding of the relative importance of various factors driving migration and making predictions that could be useful for conservation.

Reinforcement, Rationality, and Intentions: How Robust Is Automatic Reinforcement Learning in Economic Decision Making?

AbstractReinforcement learning is often observed in economic decision making and may lead to detrimental decisions. Because of its automaticity, it is difficult to avoid. In three experimental studies, we investigated whether this process could be controlled by goal intentions and implementation intentions. Participants' decisions were investigated in a probability‐updating task in which the normative rule to maximize expected payoff (Bayes' rule) conflicted with the reinforcement heuristic as a simple decision rule. Some participants were asked to set goal intentions designated to foster the optimization of rational decision making, while other participants were asked to furnish these goal intentions with implementation intentions. Results showed that controlling automatic processes of reinforcement learning is possible by means of goal intentions or implementation intentions that focus decision makers on the analysis of decision feedback. Importantly, such beneficial effects were not achieved by simply instructing participants to analyze the feedback, without defining a goal as the desired end state from a first‐person perspective. Regarding intentions supposed to shut down reinforcement processes by controlling negative affect, effects were more complex and depended on the specified goal‐directed behavior. The goal intention to suppress the disappointment elicited by negative feedback was not effective in controlling reinforcement processes. Furnishing this goal with an implementation intention even backfired and strengthened unwanted reinforcement processes. In contrast, asking participants to keep cool in response to negative decision outcomes through the use of goal intentions or implementation intentions increased decisions in line with Bayes' rule. Copyright © 2017 John Wiley & Sons, Ltd.

GD05: Careful Anticoagulation Review in Emergency Medicine (CARe-EM)

Background: The number of patients prescribed anticoagulation for stroke prevention is increasing, along with the proportion of emergency department (ED) patients who are anticoagulant users. Bleeding is the most common side effect. Inappropriate dosing, co-prescription of anti-inflammatories or aspirin, and renal impairment all increase the bleeding risk. An ED visit is an opportunity to review anticoagulant bleeding risks and intervene to prevent bleeding in patients at high risk. Objectives: To establish the 12-month incidence of bleeding in anticoagulated patients visiting the ED, to develop an ED specific anticoagulant-associated bleeding prediction score, to evaluate the ED utility of existing prediction scores. Methods: Research ethics board approval has been granted. Patients will be identified in Hamilton General and Juravinski EDs. Each patient will be followed forward in time for 12 months to document bleeding events. Population: Inclusion criteria: ED patients prescribed warfarin, rivaroxaban, dabigatran, apixaban, edoxaban or low molecular weight heparin (prevalent users). Exclusion criteria: Patients under 16 years of age. Primary outcome: The incidence of major bleeding (defined by ISTH criteria) within 12 months from the index ED visit. Secondary outcomes: Derivation of an ED prediction score to identify patients at high risk of anticoagulant-associated bleeding within 12 months. Tertiary outcomes: Evaluation of ATRIA, modified HAS-BLED and HEMORR2HAGES scores utility in predicting bleeding within 12 months. Data management: The data will be stored anonymously and securely on RedCAP. A literature search/expert discussion has identified multiple potential risk factors for bleeding. This data is collected at the time of the index ED presentation. A committee of emergency, thrombosis, gastroenterology and cardiology physicians will review each major bleeding case. Analysis: Primary analysis: a multiple logistic regression analysis to identify variables associated with major bleeding diagnosed within 12 months of the index presentation. Using the model β coefficients we will derive a simple clinical decision rule. Secondary analysis: assessing the area under the curve and optimal cut points for pre-existing bleeding prediction scores for predicting major bleeding within 12 months. Sample size calculation: With 3000 patients we expect 2700 to be anticoagulated long term, and at least 135/2700 patients will have a major bleed. This is a sufficient number for multivariate analysis to establish a simple model. We estimate 20,000 anticoagulated ED patient attendances/year. Importance: This is the first study to consider the ED visit an opportunity to prevent bleeding. We will establish a method to identify ED patients at high risk of anticoagulant-associated bleeding.

Evaluation of exchange rate point and density forecasts: An application to Brazil

This paper constructs multi-step-ahead point and density forecasts of the exchange rate. The approaches considered vary from statistical to economics-driven models, using financial and macroeconomic data and adopting either parametric or nonparametric distributions. We employ a range of statistical tools from different strands of the literature to identify which models work in practice, in terms of forecasting accuracies across different data frequencies and forecasting horizons. We propose a novel full-density/local analysis approach for collecting the many test results, and deploy a simple risk-based decision rule for ranking models. An empirical exercise with Brazilian daily and monthly data reveals that macro fundamentals are important when modeling the risk of exchange rate appreciation, whereas models that use survey information or financial data are the best way to account for the depreciation risk. These findings are relevant for econometricians, risk managers or policymakers who are interested in evaluating the accuracy of competing exchange rate models.

Intelligent hybrid system for dark spot detection using SAR data

Synthetic Aperture Radars (SAR) are the main instrument used to support oil detection systems. In the microwave spectrum, oil slicks are identified as dark spots, regions with low backscatter at sea surface. Automatic and semi-automatic systems were developed to minimize processing time, the occurrence of false alarms and the subjectivity of human interpretation. This study presents an intelligent hybrid system, which integrates automatic and semi-automatic procedures to detect dark spots, in six steps: (I) SAR pre-processing; (II) Image segmentation; (III) Feature extraction and selection; (IV) Automatic clustering analysis; (V) Decision rules and, if needed; (VI) Semi-automatic processing. The results proved that the feature selection is essential to improve the detection capability, keeping only five pattern features to automate the clustering procedure. The semi-automatic method gave back more accurate geometries. The automatic approach erred more including regions, increasing the dark spots area, while the semi-automatic method erred more excluding regions. For well-defined and contrasted dark spots, the performance of the automatic and the semi-automatic methods is equivalent. However, the fully automatic method did not provide acceptable geometries in all cases. For these cases, the intelligent hybrid system was validated, integrating the semi-automatic approach, using compact and simple decision rules to request human intervention when needed. This approach allows for the combining of benefits from each approach, ensuring the quality of the classification when fully automatic procedures are not satisfactory.

Parts-to-picker based order processing in a rack-moving mobile robots environment

This paper treats a special parts-to-picker based order processing system, where mobile robots hoist racks and bring them directly to stationary pickers. This technological innovation – known as the Kiva system – heavily influences all traditional planning problems to be solved when operating a warehouse. We, specifically, tackle the order processing in a picking station, i.e., the batching and sequencing of picking orders and the interdependent sequencing of the racks brought to a station. We formalize the resulting decision problem and provide suited solution procedures. In a comprehensive computational study we show that an optimized order picking allows to more than halve the fleet of robots compared to simple decision rules often applied in real-world warehouses.

Working dogs cooperate among one another by generalised reciprocity

Cooperation by generalised reciprocity implies that individuals apply the decision rule “help anyone if helped by someone”. This mechanism has been shown to generate evolutionarily stable levels of cooperation, but as yet it is unclear how widely this cooperation mechanism is applied among animals. Dogs (Canis familiaris) are highly social animals with considerable cognitive potential and the ability to differentiate between individual social partners. But although dogs can solve complex problems, they may use simple rules for behavioural decisions. Here we show that dogs trained in an instrumental cooperative task to provide food to a social partner help conspecifics more often after receiving help from a dog before. Remarkably, in so doing they show no distinction between partners that had helped them before and completely unfamiliar conspecifics. Apparently, dogs use the simple decision rule characterizing generalised reciprocity, although they are probably capable of using the more complex decision rule of direct reciprocity: “help someone who has helped you”. However, generalized reciprocity involves lower information processing costs and is therefore a cheaper cooperation strategy. Our results imply that generalised reciprocity might be applied more commonly than direct reciprocity also in other mutually cooperating animals.

A Novel Hybrid Learning Achievement Prediction Model: A Case Study in Gamification Education Applications (APPs)

Adaptive Neuro Fuzzy Inference System (ANFIS) used to be applied to finance, engineering, material design, and decision-making management in past research, but seldom to predict educational learning performance. In recent research, gamification learning material design is often applied to reinforce learning performance, while the prediction of gamification learning performance is seldom discussed. This study therefore applies Rough set theory to extract Core Set and generating rule, ANFIS for learning achievement predication. In order to evaluate the performance of proposed model, the VCCSEGLS dataset are collected as experimental dataset and compared with other models. The results show that the proposed method outperforms the listing models in accuracy. The three key factors are extract, (G7) Time spent on game-based learning, (L1) Examination, normal drugs and treatment, and (L2) Integration ability (time scoring, stability scoring, strain capacity, completeness scoring).The proposed model also can offer accurate predictions and provide some simple decision rules, which can be accurately used by decision-makers and game designers.

Simple Rules for Complex Decisions

From doctors diagnosing patients to judges setting bail, experts often base their decisions on experience and intuition rather than on statistical models. While understandable, relying on intuition over models has often been found to result in inferior outcomes. Here we present a new method-select-regress-and-round-for constructing simple rules that perform well for complex decisions. These rules take the form of a weighted checklist, can be applied mentally, and nonetheless rival the performance of modern machine learning algorithms. Our method for creating these rules is itself simple, and can be carried out by practitioners with basic statistics knowledge. We demonstrate this technique with a detailed case study of judicial decisions to release or detain defendants while they await trial. In this application, as in many policy settings, the effects of proposed decision rules cannot be directly observed from historical data: if a rule recommends releasing a defendant that the judge in reality detained, we do not observe what would have happened under the proposed action. We address this key counterfactual estimation problem by drawing on tools from causal inference. We find that simple rules significantly outperform judges and are on par with decisions derived from random forests trained on all available features. Generalizing to 22 varied decision-making domains, we find this basic result replicates. We conclude with an analytical framework that helps explain why these simple decision rules perform as well as they do.

Dominance-based rough fuzzy set approach and its application to rule induction

Theories of fuzzy sets and rough sets are related and complementary methodologies to handle uncertainty of vagueness and coarseness, respectively. Marrying both leads to the hybrid notion of rough fuzzy sets in order to get a more accurate account of imperfect information. In this paper, our attention is paid to ordered fuzzy decision systems, where condition criteria are preference-ordered and decision classes are not only ordered but also fuzzy. First, the dominance-based rough fuzzy approximations of an upward or downward cumulated fuzzy set are introduced in ordered fuzzy decision systems. Second, lower and upper reducts relative to a certain cumulated fuzzy set are proposed to eliminate redundant criteria in the system. Then, two approaches to attribute reduction are presented based on the discernibility matrix and the heuristic strategy, respectively. Also, decision rules are extracted directly from these approximations and some applicable and simplified decision rules are obtained according to requirements of decision makers. Finally, a case study in bankruptcy risk analysis is used to illustrate the mechanism of the proposed methods.

Co-calibration of two self-reported measures of adherence to antiretroviral therapy

ABSTRACTAdherence to antiretroviral therapy (ART) is an important determinant of clinical success assessed in many HIV studies. Harmonizing adherence data from studies that use different measures is difficult without a co-calibration equation to convert between validated instruments. Our purpose was to co-calibrate two commonly used adherence measures: the AIDS Clinical Trials Group (ACTG) questionnaire and the Visual Analog Scale (VAS). We used robust linear regression to develop a co-calibration equation in a clinical care cohort. The outcome was the 30-day VAS percentage of ART taken and the predictors were ACTG questions. We evaluated the equation’s goodness of fit in five STTR (Seek, Test, Treat, Retain) consortium studies where individuals completed both measures: 2 criminal justice; 2 international; and 1 other high-risk vulnerable population. We developed a three-phase decision rule to convert ACTG to VAS in 1045 participants. First, when the last missed dose on the ACTG was reported as >30 days ago, the VAS was set to 100% (N = 582). Second, if “doses missed” was zero for all items, VAS was 100% (N = 104). Third, among remaining participants (N = 359), VAS was estimated as 96.8% minus 2.9% times the number of missed doses (“doses per day” was non-significant). Correlation between predicted and reported VAS was r = 0.80 in the criminal justice group (N = 446), r = 0.46 in the international group (N = 311), r = 0.32 in the other vulnerable population (N = 63), and r = 0.66 overall. When outliers due to inversion of the VAS scale were excluded (n = 25), these correlations were 0.88, 0.78, 0.80, and 0.86, respectively. We concluded that a simple decision rule and equation allowed us to co-calibrate between two widely used adherence measures thus combining data from studies with different instruments. This study highlighted issues with VAS inversions and its limitations as a single item. Combining studies using different instrument facilitates larger pooled datasets to address key research questions.

Finger prick blood test could help detect serious infections.

Serious infections in children can be detected quickly by GPs using a simple decision rule and a finger prick blood test, according to a new study.

Blood test can rule out serious infections in children.

GPs can quickly detect serious infections in children using a simple decision rule and a finger prick blood test, say researchers.

Scheduling shipments in closed-loop sortation conveyors

At the very core of most automated sorting systems-- for example, at airports for baggage handling and in parcel distribution centers for sorting mail--we find closed-loop tilt tray sortation conveyors. In such a system, trays are loaded with cargo as they pass through loading stations, and are later tilted upon reaching the outbound container dedicated to a shipment's destination. This paper addresses the question of whether the simple decision rules typically applied in the real world when deciding which parcel should be loaded onto what tray are, indeed, a good choice. We formulate a short-term deterministic scheduling problem where a finite set of shipments must be loaded onto trays such that the makespan is minimized. We consider different levels of flexibility in how to arrange shipments on the feeding conveyors, and distinguish between unidirectional and bidirectional systems. In a comprehensive computational study, we compare these sophisticated optimization procedures with widespread rules of thumb, and find that the latter perform surprisingly well. For almost all problem settings, some priority rule can be identified which leads to a low-single-digit optimality gap. In addition, we systematically evaluate the performance gains promised by different sorter layouts.

Software Development Effort Estimation based-on multiple classifier system and Lines of Code

The development effort estimation is one of the most difficult problems in software project management. It is one of the most critical aspects in the early stages of the software project. Several software development effort estimation models have been proposed, however, these models are not able to obtain more than a 25 percent of accuracy, neither provide an understandable model for experts in the application area. Therefore, in this paper, we present EEpred, an explanatory model to estimate the development effort based on data of known software projects. It is a serial multiple classifier system based-on several decision trees. The model performance was evaluated by an internal validation procedure, analyzing their robustness and predictive performance. This procedure demonstrates that EEpred is able to estimate the software development effort with a 71 percent of precision. The main advantage of EEpred, regarding to other algorithms, is its ability to translate the process into a collection of simple decision rules, providing more easily interpretable knowledge that can help software engineer to improve decision-making on development planning.

What are retail investors' risk-return preferences towards renewable energy projects? A choice experiment in Germany

Citizens own nearly half the renewable energy generation capacity in Germany and have been important drivers of the country's energy transition. In contrast to citizens' important role in financing renewable energies, the energy policy and economics literature has traditionally focused on other investors, such as incumbent energy firms. To close this gap, this paper reports on a large-scale survey of 1,990 German retail investors. Conducting a choice experiment with the subset of 1,041 respondents who expressed an interest in investing in community renewable energy projects, we present a unique dataset allowing for new insights in risk-return expectations of retail investors. We find that apart from return on investment, respondents are particularly sensitive to the minimum holding period and the issuer of community renewable energy investment offerings. A minimum holding period of 10 years implies a risk premium of 2.76% points. A subsequent segmentation analysis shows that two groups of potential community renewable energy investors with different risk-return expectations can be identified: “local patriots” and “yield investors”. In contrast to professional investors, a majority of retail investors use simple decision rules such as calculating payback time or relying on their gut feeling when making investments.