Search

Scoring models support decision-making in financial institutions. Their estimation and evaluation are based on the data of previously accepted applicants with known repayment behavior. This creates sampling bias: the available labeled data offers a partial picture of the distribution of candidate borrowers, which the model is supposed to score. The paper addresses the adverse effect of sampling bias on model training and evaluation. To improve scorecard training, we propose bias-aware self-learning - a reject inference framework that augments the biased training data by inferring labels for selected rejected applications. For scorecard evaluation, we propose a Bayesian framework that extends standard accuracy measures to the biased setting and provides a reliable estimate of future scorecard performance. Extensive experiments on synthetic and real-world data confirm the superiority of our propositions over various benchmarks in predictive performance and profitability. By sensitivity analysis, we also identify boundary conditions affecting their performance. Notably, we leverage real-world data from a randomized controlled trial to assess the novel methodologies on holdout data that represent the true borrower population. Our findings confirm that reject inference is a difficult problem with modest potential to improve scorecard performance. Addressing sampling bias during scorecard evaluation is a much more promising route to improve scoring practices. For example, our results suggest a profit improvement of about eight percent, when using Bayesian evaluation to decide on acceptance rates.

This paper introduces a novel approach for efficiently distilling LLMs into smaller, application-specific models, significantly reducing operational costs and manual labor. Addressing the challenge of deploying computationally intensive LLMs in specific applications or edge devices, this technique utilizes LLMs' reasoning capabilities to generate labels and natural language rationales for unlabeled data. Our approach enhances both finetuning and distillation by employing a multi-task training framework where student models mimic these rationales alongside teacher predictions. Key contributions include the employment of zero-shot prompting to elicit teacher model rationales, reducing the necessity for handcrafted few-shot examples and lowering the overall token count required, which directly translates to cost savings given the pay-per-token billing model of major tech companies' LLM APIs. Additionally, the paper investigates the impact of explanation properties on distillation efficiency, demonstrating that minimal performance loss occurs even when rationale augmentation is not applied across the entire dataset, facilitating further reductions of tokens. This research marks a step toward the efficient training of task-specific models with minimal human intervention, offering substantial cost-savings while maintaining, or even enhancing, performance.

In many business applications, including online marketing and customer churn prevention, randomized controlled trials (RCT's) are conducted to investigate on the effect of specific treatment (coupon offers, advertisement mailings,...). Such RCT's allow for the estimation of average treatment effects as well as the training of (uplift) models for the heterogeneity of treatment effects between individuals. The problem with these RCT's is that they are costly and this cost increases with the number of individuals included into the RCT. For this reason, there is research how to conduct experiments involving a small number of individuals while still obtaining precise treatment effect estimates. We contribute to this literature a heteroskedasticity-aware stratified sampling (HS) scheme, which leverages the fact that different individuals have different noise levels in their outcome and precise treatment effect estimation requires more observations from the "high-noise" individuals than from the "low-noise" individuals. By theory as well as by empirical experiments, we demonstrate that our HS-sampling yields significantly more precise estimates of the ATE, improves uplift models and makes their evaluation more reliable compared to RCT data sampled completely randomly. Due to the relative ease of application and the significant benefits, we expect HS-sampling to be valuable in many real-world applications.

There are various applications, where companies need to decide to which individuals they should best allocate treatment. To support such decisions, uplift models are applied to predict treatment effects on an individual level. Based on the predicted treatment effects, individuals can be ranked and treatment allocation can be prioritized according to this ranking. An implicit assumption, which has not been doubted in the previous uplift modeling literature, is that this treatment prioritization approach tends to bring individuals with high treatment effects to the top and individuals with low treatment effects to the bottom of the ranking. In our research, we show that heteroskedastictity in the training data can cause a bias of the uplift model ranking: individuals with the highest treatment effects can get accumulated in large numbers at the bottom of the ranking. We explain theoretically how heteroskedasticity can bias the ranking of uplift models and show this process in a simulation and on real-world data. We argue that this problem of ranking bias due to heteroskedasticity might occur in many real-world applications and requires modification of the treatment prioritization to achieve an efficient treatment allocation., Comment: 10 pages, 4 figures

The increasing usage of new data sources and machine learning (ML) technology in credit modeling raises concerns with regards to potentially unfair decision-making that rely on protected characteristics (e.g., race, sex, age) or other socio-economic and demographic data. The authors demonstrate the impact of such algorithmic bias in the microfinance context. Difficulties in assessing credit are disproportionately experienced among vulnerable groups, however, very little is known about inequities in credit allocation between groups defined, not only by single, but by multiple and intersecting social categories. Drawing from the intersectionality paradigm, the study examines intersectional horizontal inequities in credit access by gender, age, marital status, single parent status and number of children. This paper utilizes data from the Spanish microfinance market as its context to demonstrate how pluralistic realities and intersectional identities can shape patterns of credit allocation when using automated decision-making systems. With ML technology being oblivious to societal good or bad, we find that a more thorough examination of intersectionality can enhance the algorithmic fairness lens to more authentically empower action for equitable outcomes and present a fairer path forward. We demonstrate that while on a high-level, fairness may exist superficially, unfairness can exacerbate at lower levels given combinatorial effects; in other words, the core fairness problem may be more complicated than current literature demonstrates. We find that in addition to legally protected characteristics, sensitive attributes such as single parent status and number of children can result in imbalanced harm. We discuss the implications of these findings for the financial services industry.

In response to growing FinTech competition and the need for improved operational efficiency, this research focuses on understanding the potential of advanced document analytics, particularly using multimodal models, in banking processes. We perform a comprehensive analysis of the diverse banking document landscape, highlighting the opportunities for efficiency gains through automation and advanced analytics techniques in the customer business. Building on the rapidly evolving field of natural language processing (NLP), we illustrate the potential of models such as LayoutXLM, a cross-lingual, multimodal, pre-trained model, for analyzing diverse documents in the banking sector. This model performs a text token classification on German company register extracts with an overall F1 score performance of around 80\%. Our empirical evidence confirms the critical role of layout information in improving model performance and further underscores the benefits of integrating image information. Interestingly, our study shows that over 75% F1 score can be achieved with only 30% of the training data, demonstrating the efficiency of LayoutXLM. Through addressing state-of-the-art document analysis frameworks, our study aims to enhance process efficiency and demonstrate the real-world applicability and benefits of multimodal models within banking., Comment: A Preprint

We propose a novel method for predicting time-to-event in the presence of cure fractions based on flexible survivals models integrated into a deep neural network framework. Our approach allows for non-linear relationships and high-dimensional interactions between covariates and survival and is suitable for large-scale applications. Furthermore, we allow the method to incorporate an identified predictor formed of an additive decomposition of interpretable linear and non-linear effects and add an orthogonalization layer to capture potential higher dimensional interactions. We demonstrate the usefulness and computational efficiency of our method via simulations and apply it to a large portfolio of US mortgage loans. Here, we find not only a better predictive performance of our framework but also a more realistic picture of covariate effects.

Many forecasting applications have a limited distributed target variable, which is zero for most observations and positive for the remaining observations. In the econometrics literature, there is much research about statistical model building for limited distributed target variables. Especially, there are two component model approaches, where one model is build for the probability of the target to be positive and one model for the actual value of the target, given that it is positive. However, the econometric literature focuses on effect estimation and does not provide theory for predictive modeling. Nevertheless, some concepts like the two component model approach and Heckmann's sample selection correction also appear in the predictive modeling literature, without a sound theoretical foundation. In this paper, we theoretically analyze predictive modeling for limited dependent variables and derive best practices. By analyzing various real-world data sets, we also use the derived theoretical results to explain which predictive modeling approach works best on which application., Comment: 15 pages, 2 figures

There has been intensive research regarding machine learning models for predicting bankruptcy in recent years. However, the lack of interpretability limits their growth and practical implementation. This study proposes a data-driven explainable case-based reasoning (CBR) system for bankruptcy prediction. Empirical results from a comparative study show that the proposed approach performs superior to existing, alternative CBR systems and is competitive with state-of-the-art machine learning models. We also demonstrate that the asymmetrical feature similarity comparison mechanism in the proposed CBR system can effectively capture the asymmetrically distributed nature of financial attributes, such as a few companies controlling more cash than the majority, hence improving both the accuracy and explainability of predictions. In addition, we delicately examine the explainability of the CBR system in the decision-making process of bankruptcy prediction. While much research suggests a trade-off between improving prediction accuracy and explainability, our findings show a prospective research avenue in which an explainable model that thoroughly incorporates data attributes by design can reconcile the dilemma.

Estimating treatment effects is one of the most challenging and important tasks of data analysts. In many applications, like online marketing and personalized medicine, treatment needs to be allocated to the individuals where it yields a high positive treatment effect. Uplift models help select the right individuals for treatment and maximize the overall treatment effect (uplift). A major challenge in uplift modeling concerns model evaluation. Previous literature suggests methods like the Qini curve and the transformed outcome mean squared error. However, these metrics suffer from variance: their evaluations are strongly affected by random noise in the data, which renders their signals, to a certain degree, arbitrary. We theoretically analyze the variance of uplift evaluation metrics and derive possible methods of variance reduction, which are based on statistical adjustment of the outcome. We derive simple conditions under which the variance reduction methods improve the uplift evaluation metrics and empirically demonstrate their benefits on simulated and real-world data. Our paper provides strong evidence in favor of applying the suggested variance reduction procedures by default when evaluating uplift models on RCT data.

Anticipating price developments in financial markets is a topic of continued interest in forecasting. Funneled by advancements in deep learning and natural language processing (NLP) together with the availability of vast amounts of textual data in form of news articles, social media postings, etc., an increasing number of studies incorporate text-based predictors in forecasting models. We contribute to this literature by introducing weak learning, a recently proposed NLP approach to address the problem that text data is unlabeled. Without a dependent variable, it is not possible to finetune pretrained NLP models on a custom corpus. We confirm that finetuning using weak labels enhances the predictive value of text-based features and raises forecast accuracy in the context of predicting cryptocurrency returns. More fundamentally, the modeling paradigm we present, weak labeling domain-specific text and finetuning pretrained NLP models, is universally applicable in (financial) forecasting and unlocks new ways to leverage text data., Comment: 29 pages

Generative models for images have gained significant attention in computer vision and natural language processing due to their ability to generate realistic samples from complex data distributions. To leverage the advances of image-based generative models for the time series domain, we propose a two-dimensional image representation for time series, the Extended Intertemporal Return Plot (XIRP). Our approach captures the intertemporal time series dynamics in a scale-invariant and invertible way, reducing training time and improving sample quality. We benchmark synthetic XIRPs obtained by an off-the-shelf Wasserstein GAN with gradient penalty (WGAN-GP) to other image representations and models regarding similarity and predictive ability metrics. Our novel, validated image representation for time series consistently and significantly outperforms a state-of-the-art RNN-based generative model regarding predictive ability. Further, we introduce an improved stochastic inversion to substantially improve simulation quality regardless of the representation and provide the prospect of transfer potentials in other domains.

Recurrent neural networks (RNNs) are a popular choice for modeling sequential data. Modern RNN architectures assume constant time-intervals between observations. However, in many datasets (e.g. medical records) observation times are irregular and can carry important information. To address this challenge, we propose continuous recurrent units (CRUs) -- a neural architecture that can naturally handle irregular intervals between observations. The CRU assumes a hidden state, which evolves according to a linear stochastic differential equation and is integrated into an encoder-decoder framework. The recursive computations of the CRU can be derived using the continuous-discrete Kalman filter and are in closed form. The resulting recurrent architecture has temporal continuity between hidden states and a gating mechanism that can optimally integrate noisy observations. We derive an efficient parameterization scheme for the CRU that leads to a fast implementation f-CRU. We empirically study the CRU on a number of challenging datasets and find that it can interpolate irregular time series better than methods based on neural ordinary differential equations., Comment: Accepted at ICML 2022, Baltimore, Maryland

Business success in e-commerce depends on customer perceived value. A customer with high perceived value buys, returns, and recommends items. The perceived value is at risk whenever the information load harms users' shopping experience. In e-grocery, shoppers face an overwhelming number of items, the majority of which is irrelevant for the shopper. Recommender systems (RS) enable businesses to master information overload (IO) by providing users with an item ranking by relevance. Prior work proposes RS with short personalized rankings (top-k). Given large order sizes and high user heterogeneity in e-grocery, top-k RS are insufficient to diminish IO in this domain. To fill this gap and raise business performance, this paper introduces an RS with a personalized long ranking (top-N). Undertaking a randomized field experiment, the paper establishes the merit of shifting from top-k to top-N rankings. Specifically, the proposed RS reduces IO by 29.4% and lowers users' search time by 3.3 seconds per item. The field experiment also reveals a 7% uplift in revenue due to the top-N ranking. Substantial benefits for the customer and the company highlight the business value of top-N rankings as a new design requirement for recommender systems in e-grocery.

The rise of algorithmic decision-making has spawned much research on fair machine learning (ML). Financial institutions use ML for building risk scorecards that support a range of credit-related decisions. Yet, the literature on fair ML in credit scoring is scarce. The paper makes three contributions. First, we revisit statistical fairness criteria and examine their adequacy for credit scoring. Second, we catalog algorithmic options for incorporating fairness goals in the ML model development pipeline. Last, we empirically compare different fairness processors in a profit-oriented credit scoring context using real-world data. The empirical results substantiate the evaluation of fairness measures, identify suitable options to implement fair credit scoring, and clarify the profit-fairness trade-off in lending decisions. We find that multiple fairness criteria can be approximately satisfied at once and recommend separation as a proper criterion for measuring the fairness of a scorecard. We also find fair in-processors to deliver a good balance between profit and fairness and show that algorithmic discrimination can be reduced to a reasonable level at a relatively low cost. The codes corresponding to the paper are available on GitHub., Comment: Accepted to European Journal of Operational Research

Big data and business analytics are critical drivers of business and societal transformations. Uplift models support a firm's decision-making by predicting the change of a customer's behavior due to a treatment. Prior work examines models for single treatments and binary customer responses. The paper extends corresponding approaches by developing uplift models for multiple treatments and continuous outcomes. This facilitates selecting an optimal treatment from a set of alternatives and estimating treatment effects in the form of business outcomes of continuous scale. Another contribution emerges from an evaluation of an uplift model's interpretability, whereas prior studies focus almost exclusively on predictive performance. To achieve these goals, the paper develops revenue uplift models for multiple treatments based on a recently introduced algorithm for causal machine learning, the causal forest. Empirical experimentation using two real-world marketing data sets demonstrates the advantages of the proposed modeling approach over benchmarks and standard marketing practices.

Appropriate risk management is crucial to ensure the competitiveness of financial institutions and the stability of the economy. One widely used financial risk measure is Value-at-Risk (VaR). VaR estimates based on linear and parametric models can lead to biased results or even underestimation of risk due to time varying volatility, skewness and leptokurtosis of financial return series. The paper proposes a nonlinear and nonparametric framework to forecast VaR that is motivated by overcoming the disadvantages of parametric models with a purely data driven approach. Mean and volatility are modeled via support vector regression (SVR) where the volatility model is motivated by the standard generalized autoregressive conditional heteroscedasticity (GARCH) formulation. Based on this, VaR is derived by applying kernel density estimation (KDE). This approach allows for flexible tail shapes of the profit and loss distribution, adapts for a wide class of tail events and is able to capture complex structures regarding mean and volatility. The SVR-GARCH-KDE hybrid is compared to standard, exponential and threshold GARCH models coupled with different error distributions. To examine the performance in different markets, one-day-ahead and ten-days-ahead forecasts are produced for different financial indices. Model evaluation using a likelihood ratio based test framework for interval forecasts and a test for superior predictive ability indicates that the SVR-GARCH-KDE hybrid performs competitive to benchmark models and reduces potential losses especially for ten-days-ahead forecasts significantly. Especially models that are coupled with a normal distribution are systematically outperformed.

Crime prediction is crucial to criminal justice decision makers and efforts to prevent crime. The paper evaluates the explanatory and predictive value of human activity patterns derived from taxi trip, Twitter and Foursquare data. Analysis of a six-month period of crime data for New York City shows that these data sources improve predictive accuracy for property crime by 19% compared to using only demographic data. This effect is strongest when the novel features are used together, yielding new insights into crime prediction. Notably and in line with social disorganization theory, the novel features cannot improve predictions for violent crimes.

Class imbalance is a common problem in supervised learning and impedes the predictive performance of classification models. Popular countermeasures include oversampling the minority class. Standard methods like SMOTE rely on finding nearest neighbours and linear interpolations which are problematic in case of high-dimensional, complex data distributions. Generative Adversarial Networks (GANs) have been proposed as an alternative method for generating artificial minority examples as they can model complex distributions. However, prior research on GAN-based oversampling does not incorporate recent advancements from the literature on generating realistic tabular data with GANs. Previous studies also focus on numerical variables whereas categorical features are common in many business applications of classification methods such as credit scoring. The paper propoes an oversampling method based on a conditional Wasserstein GAN that can effectively model tabular datasets with numerical and categorical variables and pays special attention to the down-stream classification task through an auxiliary classifier loss. We benchmark our method against standard oversampling methods and the imbalanced baseline on seven real-world datasets. Empirical results evidence the competitiveness of GAN-based oversampling.

This study provides a formal analysis of the customer targeting problem when the cost for a marketing action depends on the customer response and proposes a framework to estimate the decision variables for campaign profit optimization. Targeting a customer is profitable if the impact and associated profit of the marketing treatment are higher than its cost. Despite the growing literature on uplift models to identify the strongest treatment-responders, no research has investigated optimal targeting when the costs of the treatment are unknown at the time of the targeting decision. Stochastic costs are ubiquitous in direct marketing and customer retention campaigns because marketing incentives are conditioned on a positive customer response. This study makes two contributions to the literature, which are evaluated on an e-commerce coupon targeting campaign. First, we formally analyze the targeting decision problem under response-dependent costs. Profit-optimal targeting requires an estimate of the treatment effect on the customer and an estimate of the customer response probability under treatment. The empirical results demonstrate that the consideration of treatment cost substantially increases campaign profit when used for customer targeting in combination with an estimate of the average or customer-level treatment effect. Second, we propose a framework to jointly estimate the treatment effect and the response probability by combining methods for causal inference with a hurdle mixture model. The proposed causal hurdle model achieves competitive campaign profit while streamlining model building. Code is available at https://github.com/Humboldt-WI/response-dependent-costs., Comment: 22 pages, 3 figures, Accepted at European Journal of Operational Research

Uplift models support decision-making in marketing campaign planning. Estimating the causal effect of a marketing treatment, an uplift model facilitates targeting communication to responsive customers and efficient allocation of marketing budgets. Research into uplift models focuses on conversion models to maximize incremental sales. The paper introduces uplift modeling strategies for maximizing incremental revenues. If customers differ in their spending behavior, revenue maximization is a more plausible business objective compared to maximizing conversions. The proposed methodology entails a transformation of the prediction target, customer-level revenues, that facilitates implementing a causal uplift model using standard machine learning algorithms. The distribution of campaign revenues is typically zero-inflated because of many non-buyers. Remedies to this modeling challenge are incorporated in the proposed revenue uplift strategies in the form of two-stage models. Empirical experiments using real-world e-commerce data confirm the merits of the proposed revenue uplift strategy over relevant alternatives including uplift models for conver-sion and recently developed causal machine learning algorithms. To quantify the degree to which improved targeting decisions raise return on marketing, the paper develops a decomposition of campaign profit. Applying the decomposition to a digital coupon targeting campaign, the paper provides evidence that revenue uplift modeling, as well as causal machine learning, can improve cam-paign profit substantially., Comment: 53 pages including online appendix

Customer scoring models are the core of scalable direct marketing. Uplift models provide an estimate of the incremental benefit from a treatment that is used for operational decision-making. Training and monitoring of uplift models require experimental data. However, the collection of data under randomized treatment assignment is costly, since random targeting deviates from an established targeting policy. To increase the cost-efficiency of experimentation and facilitate frequent data collection and model training, we introduce supervised randomization. It is a novel approach that integrates existing scoring models into randomized trials to target relevant customers, while ensuring consistent estimates of treatment effects through correction for active sample selection. An empirical Monte Carlo study shows that data collection under supervised randomization is cost-efficient, while downstream uplift models perform competitively.

Off-the-shelf machine learning algorithms for prediction such as regularized logistic regression cannot exploit the information of time-varying features without previously using an aggregation procedure of such sequential data. However, recurrent neural networks provide an alternative approach by which time-varying features can be readily used for modeling. This paper assesses the performance of neural networks for churn modeling using recency, frequency, and monetary value data from a financial services provider. Results show that RFM variables in combination with LSTM neural networks have larger top-decile lift and expected maximum profit metrics than regularized logistic regression models with commonly-used demographic variables. Moreover, we show that using the fitted probabilities from the LSTM as feature in the logistic regression increases the out-of-sample performance of the latter by 25 percent compared to a model with only static features.

Credit scoring models support loan approval decisions in the financial services industry. Lenders train these models on data from previously granted credit applications, where the borrowers' repayment behavior has been observed. This approach creates sample bias. The scoring model (i.e., classifier) is trained on accepted cases only. Applying the resulting model to screen credit applications from the population of all borrowers degrades model performance. Reject inference comprises techniques to overcome sampling bias through assigning labels to rejected cases. The paper makes two contributions. First, we propose a self-learning framework for reject inference. The framework is geared toward real-world credit scoring requirements through considering distinct training regimes for iterative labeling and model training. Second, we introduce a new measure to assess the effectiveness of reject inference strategies. Our measure leverages domain knowledge to avoid artificial labeling of rejected cases during strategy evaluation. We demonstrate this approach to offer a robust and operational assessment of reject inference strategies. Experiments on a real-world credit scoring data set confirm the superiority of the adjusted self-learning framework over regular self-learning and previous reject inference strategies. We also find strong evidence in favor of the proposed evaluation measure assessing reject inference strategies more reliably, raising the performance of the eventual credit scoring model., Comment: Preprint of the paper accepted to ECML PKDD 2019

Accurately predicting faulty software units helps practitioners target faulty units and prioritize their efforts to maintain software quality. Prior studies use machine-learning models to detect faulty software code. We revisit past studies and point out potential improvements. Our new study proposes a revised benchmarking configuration. The configuration considers many new dimensions, such as class distribution sampling, evaluation metrics, and testing procedures. The new study also includes new datasets and models. Our findings suggest that predictive accuracy is generally good. However, predictive power is heavily influenced by the evaluation metrics and testing procedure (frequentist or Bayesian approach). The classifier results depend on the software project. While it is difficult to choose the best classifier, researchers should consider different dimensions to overcome potential bias.

The paper examines the potential of deep learning to support decisions in financial risk management. We develop a deep learning model for predicting whether individual spread traders secure profits from future trades. This task embodies typical modeling challenges faced in risk and behavior forecasting. Conventional machine learning requires data that is representative of the feature-target relationship and relies on the often costly development, maintenance, and revision of handcrafted features. Consequently, modeling highly variable, heterogeneous patterns such as trader behavior is challenging. Deep learning promises a remedy. Learning hierarchical distributed representations of the data in an automatic manner (e.g. risk taking behavior), it uncovers generative features that determine the target (e.g., trader's profitability), avoids manual feature engineering, and is more robust toward change (e.g. dynamic market conditions). The results of employing a deep network for operational risk forecasting confirm the feature learning capability of deep learning, provide guidance on designing a suitable network architecture and demonstrate the superiority of deep learning over machine learning and rule-based benchmarks., Comment: Within the "equal" contribution, Yaodong Yang contributed the core deep learning algorithm along with its experimental results, and the first draft of the manuscript (including Figure 1,2,3,4,7,8,9,11, and Table 3)

Email tracking allows email senders to collect fine-grained behavior and location data on email recipients, who are uniquely identifiable via their email address. Such tracking invades user privacy in that email tracking techniques gather data without user consent or awareness. Striving to increase privacy in email communication, this paper develops a detection engine to be the core of a selective tracking blocking mechanism in the form of three contributions. First, a large collection of email newsletters is analyzed to show the wide usage of tracking over different countries, industries and time. Second, we propose a set of features geared towards the identification of tracking images under real-world conditions. Novel features are devised to be computationally feasible and efficient, generalizable and resilient towards changes in tracking infrastructure. Third, we test the predictive power of these features in a benchmarking experiment using a selection of state- of-the-art classifiers to clarify the effectiveness of model-based tracking identification. We evaluate the expected accuracy of the approach on out-of-sample data, over increasing periods of time, and when faced with unknown senders., Comment: Accepted publication, In press, European Journal of Operational Research, 2018

The paper presents a systematic review of state-of-the-art approaches to identify patient cohorts using electronic health records. It gives a comprehensive overview of the most commonly de-tected phenotypes and its underlying data sets. Special attention is given to preprocessing of in-put data and the different modeling approaches. The literature review confirms natural language processing to be a promising approach for electronic phenotyping. However, accessibility and lack of natural language process standards for medical texts remain a challenge. Future research should develop such standards and further investigate which machine learning approaches are best suited to which type of medical data.

Leasing is a popular channel to market new cars. Pricing a leasing contract is complicated because the leasing rate embodies an expectation of the residual value of the car after contract expiration. To aid lessors in their pricing decisions, the paper develops resale price forecasting models. A peculiarity of the leasing business is that forecast errors entail different costs. Identifying effective ways to address this characteristic is the main objective of the paper. More specifically, the paper contributes to the literature through i) consolidating and integrating previous work in forecasting with asymmetric cost of error functions, ii) systematically evaluating previous approaches and comparing them to a new approach, and iii) demonstrating that forecasting with asymmetric cost of error functions enhances the quality of decision support in car leasing. For example, under the assumption that the costs of overestimating resale prices is twice that of the opposite error, incorporating corresponding cost asymmetry into forecast model development reduces decision costs by about eight percent, compared to a standard forecasting model. Higher asymmetry produces even larger improvements.