1. the complete title of one (or more) paper(s) published in the open literature describing the work that the author claims describes a human-competitive result; Towards Explainable Real Estate Valuation via Evolutionary Algorithms 2. the name, complete physical mailing address, e-mail address, and phone number of EACH author of EACH paper(s); Sebastian Angrick Hasso-Plattner-Institut University of Potsdam Prof.-Dr.-Helmert-Str. 2--3 14482 Potsdam Germany sebastian.angrick@student.hpi.de +49 331 5509-414 Ben Bals Hasso-Plattner-Institut University of Potsdam Prof.-Dr.-Helmert-Str. 2--3 14482 Potsdam Germany ben.bals@student.hpi.de +49 331 5509-414 Niko Hastrich Hasso-Plattner-Institut University of Potsdam Prof.-Dr.-Helmert-Str. 2--3 14482 Potsdam Germany niko.hastrich@student.hpi.de +49 331 5509-414 Maximilian Kleissl Hasso-Plattner-Institut University of Potsdam Prof.-Dr.-Helmert-Str. 2--3 14482 Potsdam Germany maximilian.kleissl@student.hpi.de +49 331 5509-414 Jonas Schmidt Hasso-Plattner-Institut University of Potsdam Prof.-Dr.-Helmert-Str. 2--3 14482 Potsdam Germany jonas.schmidt@student.hpi.de +49 331 5509-414 Vanja Doskoč Hasso-Plattner-Institut University of Potsdam Prof.-Dr.-Helmert-Str. 2--3 14482 Potsdam Germany vanja.doskoc@hpi.de +49 331 5509-4835 Maximilian Katzmann Institute of Theoretical Informatics Karlsruhe Institute of Technology (KIT) Am Fasanengarten 5 76131 Karlsruhe Germany maximilian.katzmann@kit.edu +49 721 608-43985 Louise Molitor Hasso-Plattner-Institut University of Potsdam Prof.-Dr.-Helmert-Str. 2--3 14482 Potsdam Germany louise.molitor@hpi.de +49 331 5509-414 Tobias Friedrich Hasso-Plattner-Institut University of Potsdam Prof.-Dr.-Helmert-Str. 2--3 14482 Potsdam Germany tobias.friedrich@hpi.de +49 331 5509-410 3. the name of the corresponding author (i.e., the author to whom notices will be sent concerning the competition); Louise Molitor 4. the abstract of the paper(s); Human lives are increasingly influenced by algorithms, which therefore need to meet higher standards not only in accuracy but also with respect to explainability. This is especially true for high-stakes areas such as real estate valuation. Unfortunately, the methods applied there often exhibit a trade-off between accuracy and explainability. One explainable approach is case-based reasoning (CBR), where each decision is supported by specific previous cases. However, such methods can be wanting in accuracy. The unexplainable machine learning approaches are often observed to provide higher accuracy but are not scrutable in their decision-making. In this paper, we apply evolutionary algorithms (EAs) to CBR predictors in order to improve their performance. In particular, we deploy EAs to the similarity functions (used in CBR to find comparable cases), which are fitted to the data set at hand. As a consequence, we achieve higher accuracy than state-of-the-art deep neural networks (DNNs), while keeping interpretability and explainability. These results stem from our empirical evaluation on a large data set of real estate offers where we compare known similarity functions, their EA-improved counterparts, and DNNs. Surprisingly, DNNs are only on par with standard CBR techniques. However, using EA-learned similarity functions does yield an improved performance. 5. a list containing one or more of the eight letters (A, B, C, D, E, F, G, or H) that correspond to the criteria (see above) that the author claims that the work satisfies; (B) The result is equal to or better than a result that was accepted as a new scientific result at the time when it was published in a peer-reviewed scientific journal. (E) The result is equal to or better than the most recent human-created solution to a long-standing problem for which there has been a succession of increasingly better human-created solutions. 6. a statement stating why the result satisfies the criteria that the contestant claims (see examples of statements of human-competitiveness as a guide to aid in constructing this part of the submission); (B) The result is equal to or better than a result that was accepted as a new scientific result at the time when it was published in a peer-reviewed scientific journal. Our paper compares our results with state-of-the-art TabNet, a high-performance and interpretable canonical deep tabular data learning architecture, and two further architectures suggested in the context of real estate proposed on Kaggle, a data science community offering machine learning competitions. We show that our approach outperforms all of the considered DNNs. Moreover, if we use at most 10 comparison objects to compute the final prediction, we obtain the same results with only a slight increase in the standard deviation. Thus, we obtain a solid human verifiability with practically no loss in performance while DNNs are black boxes, which fail the trust requirement as they only provide an opaque output without any reasoning. (E) The result is equal to or better than the most recent human-created solution to a long-standing problem for which there has been a succession of increasingly better human-created solutions. For humans it is hard to define the similarity of two properties. Is a 3-room apartment with a total area of 95m² more similiar to a 2-room apartment with 100m² or to a 4-room apartment with 90m²? Many years of professional experience are necessary to develop some sensibility. Even today in many banks most of the real estate experts estimate the price of a property on the basis of their personal similar past experiences. This takes a lot of time, and for non-experts the estimations of properties are non-comprehensible in many cases. Our approach supports the valuation of properties by providing similarity functions, found via EAs. While previous approaches do not scale to large data sets our approach reduces the complexity for a single prediction while keeping its explainable characteristics by providing similiar objects. Further, since our approach makes explicit connections between properties, it is able to evaluate the location of a property (one of its most important aspects) using surrounding properties. In contrast, DNNs seem to be unable to make these relationships. 7. a full citation of the paper (that is, author names; publication date; name of journal, conference, technical report, thesis, book, or book chapter; name of editors, if applicable, of the journal or edited book; publisher name; publisher city; page numbers, if applicable); Author names: Sebastian Angrick, Ben Bals, Niko Hastrich, Maximilian Kleissl, Jonas Schmidt, Vanja Doskoč, Maximilian Katzmann, Louise Molitor, Tobias Friedrich Name of the conference: The Genetic and Evolutionary Computation Conference 2022 (GECCO 2022) 8. a statement either that "any prize money, if any, is to be divided equally among the co-authors" OR a specific percentage breakdown as to how the prize money, if any, is to be divided among the co-authors; Any prize money, if any, is to be divided equally among the three undergraduate students of this paper, who worked on the paper as an extension of their bachelor project. The students' names are: – Sebastian Angrick – Ben Bals – Jonas Schmidt 9. a statement stating why the authors expect that their entry would be the "best," and The real estate market has many profit opportunities. These are influenced by several factors, such as location, furnishing, and the overall market situation. People and companies are drawn to this market without being real estate experts. However, setting the price of a house or apartement correctly is of crucial importance to buyer, seller, and the bank providing financing. We believe that our solution tackles this highly important problem elegantly, due to the following properties: – It is explainable. In contrast to black box approaches like DNNs, we provide an explainable approach. Even though the EA itself is non-explainable, the resulting similarity function can be interpreted. Moreover, the provided comparison properties provide a reasoning and allow a rapid integration into existing, human-performed real estate valuation. – It has high accuracy. Our EA-improved CBR approach consistenly yielded the best results with an average MAPE of 12.1% with a standard deviation of 0.1%. – It can deal with large data sets. In contrast to previous approaches, which rely on maximum likelihood estimation and numerical approaches, our approach reduces the complexity for a single prediction from linear to logarithmic. 10. An indication of the general type of genetic or evolutionary computation used, such as GA (genetic algorithms), GP (genetic programming), ES (evolution strategies), EP (evolutionary programming), LCS (learning classifier systems), GE (grammatical evolution), GEP (gene expression programming), DE (differential evolution), etc. GA (genetic algorithms) 11. The date of publication of each paper. If the date of publication is not on or before the deadline for submission, but instead, the paper has been unconditionally accepted for publication and is “in press” by the deadline for this competition, the entry must include a copy of the documentation establishing that the paper meets the “in press” requirement. Copy of the paper attached to the e-mail. To be published in Proceedings, GECCO 2022, Boston