PEC-PG. CNPq / As cooperativas de crédito são regidas pelos mesmos princípios cooperativistas, mas agregam funções financeiras como os bancos tradicionais. A solidariedade na área financeira permite que as cooperativas de crédito levem aos seus cooperados os fundos poupados ou repassando os de desenvolvimento governamental das políticas públicas em forma de crédito. É evidente, no entanto, que nem sempre os “apoios” são devolvidos da mesma maneira como foram recebidos, neste mundo de muita perfeição e de alta competitividade. Assim, integra o elemento estranho no meio onde se imagina seu difícil acesso, a falta de confiança no cumprimento do contrato de crédito contraído pelos credores (cooperados). É conhecido como Risco de Crédito ou de um eventual não pagamento das dívidas contraídas pelos cooperados, o principal fator para processo de Gestão de Risco de Crédito nas instituições cooperativas do mundo. Desse modo, se insere o problema da pesquisa, sendo nova a gestão de risco nas cooperativas de crédito e num mercado assombrado pelas crises financeiras, e com isso se questiona: em que medida ou em que ponto o cumprimento das normas da gestão de risco de crédito nessas instituições vem sendo atendido? Com base nessa preocupação, traçaram-se os objetivos, que apontam em avaliar as práticas de gestão de risco em cooperativas de crédito do Sudoeste do Paraná (PR) de acordo com as resoluções do Conselho Monetário Nacional (CMN). Para isso, tem-se como objetivos específicos: listar as práticas de gestão de risco necessárias para a concessão de crédito; identificar de que forma estão sendo abordadas as práticas de gestão de risco de crédito, considerando-se normas de CMN; verificar qual é o impacto da gestão de risco percebido pelos gestores nas cooperativas de crédito; levantar em termos quantitativos quanto essas cooperativas perdem anualmente em inadimplência (default); comparar as práticas das cooperativas pesquisadas entre si e com as resoluções do CMN. O trabalho se justifica pela importância do processo de gestão de risco para as instituições cooperativas, para a academia e pela contribuição no reforço àquilo que o CMN vem recomendado nas suas resoluções. Para se atingir os resultados pretendidos, foi aplicado um estudo de caso múltiplo nas cooperativas CRESOL, SICOOB e SICREDI/PR, com abordagem qualitativa e quantitativa, em parte com dados secundários e primários. Os dados secundários foram levantados por meio da revisão da literatura referente às áreas de cooperativismo em geral e de crédito, a gestão de risco de crédito e normas e as resoluções do CMN sobre o tema, além de informações nos Relatórios Financeiros das três cooperativas. Os dados primários foram levantados por meio de aplicação de questionário nessas instituições, sendo entrevistados os Gerentes de Crédito, o Assessor Sênior da Controladoria e o Assessor e Supervisor de Crédito. Os dados levantados foram comparados entre as informações dos entrevistados da mesma instituição e depois com as outras, sendo que, na análise dos dados, os nomes das mesmas (instituições) deixaram de ser citadas e foi atribuído o nome de CASO (CASO1, CASO2, e 3 ). Foi identificada uma preocupação com a gestão de risco nessas instituições. Também se percebeu que o crédito só é liberado quando passa por análise de, no mínimo, três pessoas. Existem procedimentos de análise de crédito que seguem todas as alçadas necessárias para que a proposta seja deferida e também são avaliados de acordo com a renda do proponente e as cooperativas singulares são classificadas de acordo com seus patrimônios de Referências (PR) e Patrimônios de Referências exigidas (PRE), para receberem crédito das suas centrais e até para liberarem crédito aos seus cooperados Pessoa Jurídica, caso existam. Mesmo assim, foram localizadas as perdas (prejuízos), algumas mais acentuadas do que outras. Percebe-se que ainda existem reduzidos estudos sobre gestão de risco nas cooperativas de crédito, dada à situação em que se encontravam as mesmas, mas atualmente vale reforçar as pesquisas para conhecer como essas instituições lidam com a gestão de risco de crédito e outros riscos. / Credit unions are governed by the same cooperative principles but add financial functions as traditional banks. Solidarity in the financial area allows credit unions to take to their members or transferring funds spared government development of public policy in the form of credit. But it is not always clear that the "support" are returned the same way they were received, in this world of too much perfection and high competitiveness. Integrates the foreign element in the middle where you think its difficult access, lack of confidence in the fulfillment of the credit agreement contracted by creditors (cooperative). Known as credit risk or a possible non-payment of debts contracted by the cooperatives the main factor for Credit Risk Management process in cooperative institutions in the world. Thus falls the research problem, and new risk management in credit unions and a market haunted by financial crises, and if it asks: To what extent or at what point the compliance of the management of credit risk these institutions has been met? Based on this concern were traced objectives, which aim to assess the practices of risk management in credit unions Southwest of Paraná (PR) according to the resolutions of the National Monetary Council (CMN). For this, one has the following specific objectives: List the risk management practices needed to grant credit; Identify how they are being addressed management practices of credit risk considering CMN standard; Ascertain the impact of the management of risk perceived by managers in credit unions; Rise in quantitative terms as these cooperatives lost annually in default; Compare practices of cooperatives surveyed each other and with the resolutions of the CMN. The work is justified by the importance of the risk management process for cooperative institutions, academia and the contribution to the reinforcement to what the CMN has been recommended in its resolutions. And to achieve the desired results a study of multiple case was applied in cooperative CRESOL, SICOOB and SICREDI / PR, with qualitative and quantitative approach in part with secondary and primary data. The secondary data were collected by reviewing the literature pertaining to the areas of cooperative movement in general and credit risk management and credit standards and CMN resolutions on the topic, also collected in the Financial Reports of the three Credit unions. Primary data were collected through a questionnaire in these institutions, being interviewed Managers of Credit, Senior Advisor of the Comptroller, Assessor and Supervisor of Credit. And the data obtained were compared to the information of the respondents from the same institution and then with the other, in the data analysis of the same names (institutions) are no longer quoted and was assigned the name of CASE (case1 case2 and 3). A concern with risk management in these institutions was identified. Also noticed that credit is only been released when passing by analysis of at least three people. There are procedures for credit analysis that follows all the necessary limits for the proposal to be granted and are evaluated according to the income of the applicant and the individual cooperatives are classified according to their Wealth of References (PR in Portuguese) and Heritage References Required (PRE in Portuguese), to receive credit of their plants and up to release credit to their members Corporations if any. Yet losses were located some steeper others do not. Realize that there are still smaller studies on risk management in credit unions, given the situation they were in the same, but is currently worth strengthen research to know how these institutions deal with the management of credit risk and other risks. / 5000

Factoring innebär försäljning av fakturor till tredjepart och därmed möjlighet att få in kapital snabbt och har blivit alltmer populärt bland företag idag. Ett fakturaköp innebär en viss kreditrisk för företaget i de fall som fakturan inte blir betald och som köpare av kapital önskar man att minimera den risken. Aros Kapital erbjuder sina kunder tjänsten factoring. Under detta projekt undersöks möjligheten att använda maskininlärningsmetoder för att bedöma om en faktura är en bra eller dålig investering. Om maskininlärningen visar sig vara bättre än manuell hantering kan även bättre resultat uppnås i form av minskade kreditförluster, köp av fler fakturor och därmed ökad vinst. Fyra maskininlärningsmetoder jämfördes: beslutsträd, slumpmässig skog, Adaboost och djupa neurala nätverk. Utöver jämförelse sinsemellan har metoderna jämförts med Aros befintliga beslut och nuvarande regelmotor. Av de jämförda maskininlärningsmetoderna presterade slumpmässig skog bäst och visade sig bättre än Aros befintliga beslut på de testade fakturorna, slumpmässig skog fick F1-poängen 0,35 och Aros 0,22 . / Today, companies can sell their invoices to a third party in order to to quickly capitalize them. This is called factoring. For the financial institute which serve as the third party, the purchase of an invoice infers a certain risk in case the invoice is not paid, a risk the financial institute would like to minimize. Aros Kapital is a financial institute that offers factoring as one of their services. This project at Aros Kapital evaluated the possibility of using machine learning to determine whether or not an invoice will be good investment for the financial institute. If the machine learning algorithm performs better than manual handling and by minimizing credit losses and buying more invoices this could lead to an increase in profit for Aros. Four machine learning algorithms have been compared: decision trees, random forest, Adaboost and deep neural network. Beyond the comparison between the four algorithms, the algorithms were also compared with Aros actual decision and Aros current rule engine solution. The  results show that random forest is the best performing algorithm and it also shows a slight improvement on performance compared to Aros actual decision, random forest got an F1- core of 0.35 and Aros 0.22.

The aim is to develop a new decision-making model for credit-loans. The model will be specific for credit applicants of the OKQ8 bank, becauseit is based on data of earlier applicants of credit from the client (the bank). The final model is, in effect, functional enough to use informationabout a new applicant as input, and predict the outcome to either the good risk group or the bad risk group based on the applicant’s properties.The prediction may then lay the foundation for the decision to grant or deny credit loan. Because of the skewed distribution in the response variable, different sampling techniques are evaluated. These include oversampling with SMOTE, random undersampling and pure oversampling in the form of scalar weighting of the minority class. It is shown that the predictivequality of a classifier is affected by the distribution of the response, and that the oversampled information is not too redundant. Three classification techniques are evaluated. Our results suggest that a multi-layer neural network with 18 neurons in a hidden layer, equippedwith an ensemble technique called boosting, gives the best predictive power. The most successful model is based on a feed forward structure andtrained with a variant of back-propagation using conjugate-gradient optimization. Two other models with a good prediction quality are developed using logistic regression and a decision tree classifier, but they do not reach thelevel of the network. However, the results of these models are used to answer the question regarding which customer properties are importantwhen determining credit risk. Two examples of important customer properties are income and the number of earlier credit reports of the applicant. Finally, we use the best classification model to predict the outcome of a set of applicants declined by the existent filter. The results show that thenetwork model accepts over 60 % of the applicants who had previously been denied credit. This may indicate that the client’s suspicionsregarding that the existing model is too restrictive, in fact are true.

信用風險或信用違約意旨金融機構提供給客戶服務卻未得償還的機率，故其在銀行信貸決策的領域是常被鑽研的對象，因為其對於金融機構所扮演的角色尤其重要，對商業銀行來說更是常難以解釋或控制，然而拜現今進步的科技所賜，金融機構可以藉由操控較過去低的成本即可進一步發展強健且精煉的系統與模型去做預測還有信用風險的控管，有鑑於對客戶的評分自大數據時代來臨起，即使是學生亦開始有了可以評鑑的痕跡，憑藉前人所實驗或仰賴的基本考量面向如客戶基本資料、財力狀況或是其於該公司今昔的借貸訊息，再輔以藉由開放資料所帶來的資訊，發想可能影響信用違約率的變數如外在規範對該客戶的紀錄，想驗證是否真有尚可開發的方向，若有則其影響可以到多深。 眾所皆知從過去到現在即有很多種方法被開創以及提出以預測信用違約率，當然所使用的方法和金融機構本身的複雜性、規模大小以及信貸類型有關，最常見的有判別分析，但其對於變數有嚴格的假設，而新興的方法神經網路可以克服判別分析的缺陷且預測的效能也不錯，但神經網路只給予預測結果而運算過程是未知的，對於想要了解變數間的關係無濟於事，故還是選擇從可以對二元分類做預測亦可以藉由模型係數看到應變數和自變數間關係的羅吉斯迴歸方法著手，而研究過程即是依著前人對於羅吉斯迴歸在信用風險上的繩索摸索，將資料如何清理、變數如何轉換、模型如何建立以及最後如何篩選做一個完整的陳述，縱然長道漫漫，對於研究假設在結果終得驗證也始見曙光，考慮的新面向確有其影響力，而在模型係數上也看到其影響的大小，為了更彰顯羅吉斯迴歸對於變數間提供的訊息，故在最後將研究結果以較文字易讀的視覺化方式作呈現。 / Credit risk or credit default means the probability of non-repayment that banks or financial institutions get after they provide services to their customers. Credit risk is also studied intensively in the field of bank lending strategy because it’s usually hard to interpret and control. However, thanks to advanced technology nowadays, banks can manipulate reduced cost to develop robust and well-trained system and models so as to predict and mange credit risk. In the light of the score on customers from the beginning of big data era, every single one can be tracked to assess even though he or she is student. Relying on common facets like personal information, financial statement and past relationship of loan in a specific bank, come up with possible variables like regulations which influence credit risk according to information from open data. Try to verify if there is a new aspect of modeling and how far it effects. As everyone knows, there are several created and offered methodologies in order to predict credit default. They differ from complexity of banks and institutions, size and type of loan. One of the most popular method is discriminant analysis, but variables are restricted to its assumption. Neural network can fix the flaws of the assumption and work efficiently. Considering the unknown process of calculation in neural network, choose logistic regression as research method which can see the relationship between variables and predict the binary category. With the posterior research on credit risk, make a complete statement about how to clean data, how to transform variables and how to build or screen models. Although the procedure is complicated, the result of this study still validates original hypothesis that new aspect indeed has an impact on credit risk and the coefficient shows how deep it affects.

Nowadays, the use of credit scoring models in the financial sector is a common practice. Credit scoring plays an important role in profitability and transparency of lending business. Given the high credit volumes, even a small improvement of discriminatory and predictive power of a credit scoring model may provide a substantial additional profit. Scoring models are applied on the through-the-door population, however, for creating them or adjusting already existing credit rules, it is usual to use only the data corresponding to accepted applicants for which payment discipline can be observed. This discrepancy can lead to reject bias (or selection bias in general). Methods trying to eliminate or reduce this phenomenon are known by the term reject inference. In general, these methods try to assess the behavior of rejected applicants or to obtain an additional information about them. In the dissertation thesis, I dealt with the enlargement method which is based on a random acceptance of applicants that would have been rejected. This method is not only time consuming but also expensive. Therefore I looked for the ways how to reduce the cost of acquiring additional information about rejected applicants. As a result, I have proposed a modification which I called the enlargement method with sorting variable. It was validated on real bank database with two possible sorting variables and the results were compared with the original version of the method. It was shown that both tested approaches can reduce its cost while retaining the accuracy of the scoring models.

This project will explore machine learning approaches that are used in creditscoring. In this study we consider consumer credit scoring instead of corporatecredit scoring and our focus is on methods that are currently used in practiceby banks such as logistic regression and decision trees and also compare theirperformance against machine learning approaches such as support vector machines (SVM), neural networks and random forests. In our models we addressimportant issues such as dataset imbalance, model overfitting and calibrationof model probabilities. The six machine learning methods we study are support vector machine, logistic regression, k-nearest neighbour, artificial neuralnetworks, decision trees and random forests. We implement these models inpython and analyse their performance on credit dataset with 30000 observations from Taiwan, extracted from the University of California Irvine (UCI)machine learning repository.

Der Aufstieg des maschinellen Lernens (ML) und die rasante Digitalisierung der Wirtschaft haben die Entscheidungsprozesse in der Finanzbranche erheblich verändert. Finanzinstitute setzen zunehmend auf ML, um die Entscheidungsfindung zu unterstützen. Kreditscoring ist eine der wichtigsten ML-Anwendungen im Finanzbereich. Die Aufgabe von Kreditscoring ist die Unterscheidung ob ein Antragsteller einen Kredit zurückzahlen wird. Finanzinstitute verwenden ML, um Scorecards zu entwickeln, die die Ausfallwahrscheinlichkeit eines Kreditnehmers einschätzen und Genehmigungsentscheidungen automatisieren. Diese Dissertation konzentriert sich auf drei große Herausforderungen, die mit dem Aufbau von ML-basierten Scorekarten für die Bewertung von Verbraucherkrediten verbunden sind: (i) Optimierung von Datenerfassungs- und -speicherkosten bei hochdimensionalen Daten von Kreditantragstellern; (ii) Bewältigung der negativen Auswirkungen von Stichprobenverzerrungen auf das Training und die Bewertung von Scorekarten; (iii) Messung und Sicherstellung der Fairness von Instrumenten bei gleichzeitig hoher Rentabilität. Die Arbeit bietet und testet eine Reihe von Instrumenten, um jede dieser Herausforderungen zu lösen und die Entscheidungsfindung in Finanzinstituten zu verbessern. Erstens entwickeln wir Strategien zur Auswahl von Merkmalen, die mehrere unternehmensbezogene Zielfunktionen optimieren. Unsere Vorschläge reduzieren die Kosten der Datenerfassung und verbessern die Rentabilität der Modelle. Zweitens schlagen wir Methoden zur Abschwächung der negativen Auswirkungen von Stichprobenverzerrungen vor. Unsere Vorschläge gleichen die Verluste aufgrund von Verzerrungen teilweise aus und liefern zuverlässigere Schätzungen der künftigen Scorecard-Leistung. Drittens untersucht die Arbeit faire ML-Praktiken in Kreditscoring. Wir katalogisieren geeignete algorithmische Optionen für die Einbeziehung von Fairness-Zielen und verdeutlichen den Kompromiss zwischen Gewinn und Fairness. / The rise of machine learning (ML) and the rapid digitization of the economy has substantially changed decision processes in the financial industry. Financial institutions increasingly rely on ML to support decision-making. Credit scoring is one of the prominent ML applications in finance. The task of credit scoring is to distinguish between applicants who will pay back the loan or default. Financial institutions use ML to develop scoring models to estimate a borrower's probability of default and automate approval decisions. This dissertation focuses on three major challenges associated with building ML-based scorecards in consumer credit scoring: (i) optimizing data acquisition and storage costs when dealing with high-dimensional data of loan applicants; (ii) addressing the adverse effects of sampling bias on training and evaluation of scoring models; (iii) measuring and ensuring the scorecard fairness while maintaining high profitability. The thesis offers a set of tools to remedy each of these challenges and improve decision-making practices in financial institutions. First, we develop feature selection strategies that optimize multiple business-inspired objectives. Our propositions reduce data acquisition costs and improve model profitability and interpretability. Second, the thesis illustrates the adverse effects of sampling bias on model training and evaluation and suggests novel bias correction frameworks. The proposed methods partly recover the loss due to bias, provide more reliable estimates of the future scorecard performance and increase the resulting model profitability. Third, the thesis investigates fair ML practices in consumer credit scoring. We catalog algorithmic options for incorporating fairness goals in the model development pipeline and perform empirical experiments to clarify the profit-fairness trade-off in lending decisions and identify suitable options to implement fair credit scoring and measure the scorecard fairness.

This study reports the development of an instrument intended to measure mathematics teachers' knowledge of Cognitively Guided Instruction (CGI). CGI is a mathematics professional development framework based on how students think about and solve problems and how that knowledge guides instruction for developing mathematical understanding. The purpose of this study was to (a) analyze and revise the original CGI Teacher Knowledge Assessment (CGI TKA), (b) administer the revised CGI TKA, and (c) analyze the results from the revised CGI TKA. As part of the revision of the original CGI TKA, distractor analysis identified distractors that could be improved. Experts in CGI content were interviewed to identify ways in which the content of the CGI TKA could be improved, and some new items were created based on their feedback. Formatting changes were also made to administer the assessment electronically.After the original CGI TKA was revised, the revised CGI TKA was administered to teachers who had been trained in CGI. Two hundred thirteen examinees completed the revised CGI TKA and the results were analyzed. Exploratory and confirmatory factor analyses showed 21 of the items loaded adequately onto one factor, considered to be overall knowledge of CGI. The Rasch model was used to estimate item difficulty and person abilities as well as to compare models using dichotomous and partial credit scoring. Advantages and disadvantages of using partial credit scoring as compared to dichotomous scoring are discussed. Except under special circumstances, the dichotomous scoring produced better fitting models and more reliable scores than the partial credit scoring. The reliability of the scores was estimated using Raykov's rho coefficient. Overall, the revised CGI TKA appears to validly and reliably measure teachers' CGI knowledge.

This thesis explores how survival analysis models performs for default risk prediction of small-to-medium sized enterprises (SME) and investigates when survival analysis models are preferable to use. This is examined by comparing the performance of three deep learning models in a survival analysis setting, a traditional survival analysis model Cox Proportional Hazards, and a traditional credit risk model logistic regression. The performance is evaluated by three metrics; concordance index, integrated Brier score and ROC-AUC. The models are trained on financial data from Swedish SME holding profit and loss statement and balance sheet results. The dataset is divided into two feature sets: a smaller and a larger, additionally the features are binned.  The results show that DeepHit and Logistic Hazard performed the best with the three metrics in mind. In terms of the AUC score all three deep learning survival models generally outperform the logistic regression model. The Cox Proportional Hazards (Cox PH) showed worse performance than the logistic regression model on the non-binned feature sets while having more comparable results in the case where the data was binned. In terms of the concordance index and integrated Brier score the Cox Proportional Hazards model consistently performed the worst out of all survival models. The largest significant performance gain for the concordance index and AUC score was however seen by the Cox PH model when binning was applied to the larger feature set. The concordance index went from 0.65 to 0.75 and the test AUC went from 76.56% to 83.91% for the larger set to larger dataset with binned features. The main conclusions is that the neural networks models did outperform the traditional models slightly and that binning had a great impact on all models, but in particular for the Cox PH model. / Det här examensarbete utreder hur modeller inom överlevnadsanalys presterar för kreditriskprediktion på små och medelstora företag (SMF) och utvärderar när överlevnadsanalys modeller är att föredra. För att besvara frågan jämförs prestandan av tre modeller för djupinlärning i en överlevnadsanalysmiljö, en traditionell överlevnadsanalys modell: Cox Proportional Hazards och en traditionell kreditriskmodell: logistik regression. Prestandan har utvärderats utifrån tre metriker; concordance index, integrated Brier score och AUC. Modellerna är tränade på finansiell data från små och medelstora företag som innefattar resultaträkning och balansräkningsresultat. Datasetet är fördelat i ett mindre variabelset och ett större set, dessutom är variablerna binnade.  Resultatet visar att DeepHit och Logistic Hazard presterar bäst baserat på alla metriker. Generellt sett är AUC måttet högre för alla djupinlärningsmodeller än för den logistiska regressionen. Cox Proportional Hazards (Cox PH) modellen presterar sämre för variabelset som inte är binnade men får jämförelsebar resultat när datan är binnad. När det gäller concordance index och integrated Brier score så har Cox PH överlag sämst resultat utav alla överlevnadsmodeller. Den största signifikanta förbättringen i resultatet för concordance index och AUC ses för Cox PH när datan binnas för det stora variabelsetet. Concordance indexet gick från 0.65 till 0.75 och test AUC måttet gick från 76.56% till 83.91% för det större variabel setet till större variabel setet med binnade variabler. De huvudsakliga slutsatserna är att de neurala nätverksmodeller presterar något bättre än de traditionella modellerna och att binning är mycket gynnsam för alla modeller men framförallt för Cox PH.

The field of explainable artificial intelligence (XAI) has gained momentum in recent years following the increased use of AI systems across industries leading to bias, discrimination, and data security concerns. Several conceptual frameworks for how to reach AI systems that are fair, transparent, and understandable have been proposed, as well as a number of technical solutions improving some of these aspects in a research context. However, there is still a lack of studies examining the implementation of these concepts and techniques in practice. This research aims to bridge the gap between prominent theory within the area and practical implementation, exploring the implementation and evaluation of XAI models in the Swedish credit scoring industry, and proposes a three-step framework for the implementation of local explanations in practice. The research methods used consisted of a case study with the model development at UC AB as a subject and an experiment evaluating the consumers' levels of trust and system understanding as well as the usefulness, persuasive power, and usability of the explanation for three different explanation prototypes developed. The framework proposed was validated by the case study and highlighted a number of key challenges and trade-offs present when implementing XAI in practice. Moreover, the evaluation of the XAI prototypes showed that the majority of consumers prefers rulebased explanations, but that preferences for explanations is still dependent on the individual consumer. Recommended future research endeavors include studying a longterm XAI project in which the models can be evaluated by the open market and the combination of different XAI methods in reaching a more personalized explanation for the consumer. / Under senare år har antalet AI implementationer stadigt ökat i flera industrier. Dessa implementationer har visat flera utmaningar kring nuvarande AI system, specifikt gällande diskriminering, otydlighet och datasäkerhet vilket lett till ett intresse för förklarbar artificiell intelligens (XAI). XAI syftar till att utveckla AI system som är rättvisa, transparenta och begripliga. Flera konceptuella ramverk har introducerats för XAI som presenterar etiska såväl som politiska perspektiv och målbilder. Dessutom har tekniska metoder utvecklats som gjort framsteg mot förklarbarhet i forskningskontext. Däremot saknas det fortfarande studier som undersöker implementationer av dessa koncept och tekniker i praktiken. Denna studie syftar till att överbrygga klyftan mellan den senaste teorin inom området och praktiken genom en fallstudie av ett företag i den svenska kreditupplysningsindustrin. Detta genom att föreslå ett ramverk för implementation av lokala förklaringar i praktiken och genom att utveckla tre förklaringsprototyper. Rapporten utvärderar även prototyperna med konsumenter på följande dimensioner: tillit, systemförståelse, användbarhet och övertalningsstyrka. Det föreslagna ramverket validerades genom fallstudien och belyste ett antal utmaningar och avvägningar som förekommer när XAI system utvecklas för användning i praktiken. Utöver detta visar utvärderingen av prototyperna att majoriteten av konsumenter föredrar regelbaserade förklaringar men indikerar även att preferenser mellan konsumenter varierar. Rekommendationer för framtida forskning är dels en längre studie, vari en XAI modell introduceras på och utvärderas av den fria marknaden, dels forskning som kombinerar olika XAI metoder för att generera mer personliga förklaringar för konsumenter.