Virtual prototyping is increasingly used by businesses to streamline operations, cut costs, and enhance daily operations. This often includes a variety of modeling techniques among which, complex, black-box models. The path from model development to utilization in applied contexts is yet long. Domain experts need to be convinced of the validity of the models and to trust their predictions. To be used in the field, model capabilities need to be affordable, that is, allow rapid and interactive scenario building, even for non-experts. Complex relations governed by statistical interactions must be unveiled for users to understand unexpected predictions. We propose Interact, a model-agnostic, visual what-if tool for regression problems, supporting (1) the visualization of statistical interactions between features, (2) the creation of interactive what-if scenarios using predictive models, (3) the evaluation of model quality and building trust, and (4) the externalization of knowledge through model explainability. While the approach applies in various industrial contexts, we validate the application purpose and design with a detailed case study and a qualitative user study with engineers in the tire industry. By unraveling statistical interactions between features, the INTERACT tool proves to be useful to increase the transparency of black-box machine learning models. We also reflect on lessons learned concerning the development of visual what-if tools for virtual product development and beyond.
KwonBCChoiMJKimJT, et al. RetainVis: visual analytics with interpretable and interactive recurrent neural networks on electronic medical records. IEEE Trans Vis Comput Graph2019; 25(1): 299–309. DOI: 10.1109/TVCG.2018.2865027
2.
GotzDBorlandD.Data-driven healthcare: challenges and opportunities for interactive visualization. IEEE Comput Graph Appl2016; 36(3): 90–96.
3.
StirnbergRCermakJKotthausS, et al. Meteorology-driven variability of air pollution (PM1) revealed with explainable machine learning. Atmos Chem Phys2021; 21(5): 3919–3948.
4.
RautenhausMBottingerMSiemenS, et al. Visualization in meteorology-a survey of techniques and tools for data analysis tasks. IEEE Trans Vis Comput Graph2018; 24(12): 3268–3296.
5.
SukhijaNTatineniMBrownN, et al. Topic modeling and visualization for big data in social sciences. In 2016 Int. IEEE conf. on ubiquitous intelligence & computing, advanced and trusted computing, scalable computing and communications, cloud and big data computing, internet of people, and smart world congress, 2016, pp.1198–1205. IEEE. DOI:10.1109/UIC-ATC-ScalCom-CBDComIoP-SmartWorld.2016.0183.
ChatzimparmpasAMartinsRMJusufiI, et al. A survey of surveys on the use of visualization for interpreting machine learning models. Inf Vis2020; 19(3): 207–233.
WangX. (ed.) Automotive tire noise and vibrations: analysis, measurement and simulation. Oxford: Butterworth-Heinemann, 2020.
11.
ClarkSDodgeR.The rolling resistance of pneumatic tires. Final Report. Report, U.S. Department of Transportation, NHTSA, Office of Research and Development, 1979.
HastieTTibshiraniRFriedmanJH, et al. The elements of statistical learning: data mining, inference, and prediction. New York, NY: Springer, 2009. Vol. 2.
17.
IwaoKYamazakiI.A study on the mechanism of tire/road noise. JSAE Rev1996; 17(2): 139–144.
18.
GuidottiRMonrealeARuggieriS, et al. A survey of methods for explaining black box models. ACM Comput Surv2019; 51(5): 1–42.
19.
EndertARibarskyWTurkayC, et al. The state of the art in integrating machine learning into visual analytics. Comput Graph Forum2017; 36(8): 458–486.
20.
HohmanFMKahngMPientaR, et al. Visual analytics in deep learning: an interrogative survey for the next frontiers. IEEE Trans Vis Comput Graph2019; 25(8): 2674–2693. DOI: 10.1109/TVCG.2018.2843369
21.
YuanJChenCYangW, et al. A survey of visual analytics techniques for machine learning. Comput Vis Media2021; 7(1): 3–36.
ChatzimparmpasAMartinsRMJusufiI, et al. The state of the art in enhancing trust in machine learning models with the use of visualizations. Comput Graph Forum2020; 39(3): 713–756.
24.
AdadiABerradaM.Peeking inside the black-box: a survey on explainable artificial intelligence (XAI). IEEE Access2018; 6: 52138–52160.
25.
ZhaoXWuYLeeDL, et al. IForest: interpreting random forests via visual analytics. IEEE Trans Vis Comput Graph2019; 25(1): 407–416. DOI: 10.1109/TVCG.2018.2864475
26.
KahngMThoratNChauDHP, et al. GAN Lab: understanding complex deep generative models using interactive visual experimentation. IEEE Trans Vis Comput Graph2018; 25(1): 310–320. DOI: 10.1109/TVCG.2018.2864500
27.
WangJGouLShenHW, et al. DQNViz: A visual analytics approach to understand deep Q-networks. IEEE Trans Vis Comput Graph2018; 25(1): 288–298. DOI: 10.1109/TVCG.2018.2864504
28.
KahngMAndrewsPYKalroA, et al. ActiVis: visual exploration of industry-scale deep neural network models. IEEE Trans Vis Comput Graph2018; 24(1): 88–97.
29.
LiuMLiuSSuH, et al. Analyzing the noise robustness of deep neural networks. In: Proceedings of the 2018 IEEE conference on visual analytics science and technology. DOI: 10.1109/VAST.2018.8802509.2001.09395.
30.
MingYCaoSZhangR, et al. Understanding hidden memories of recurrent neural networks. In: Proceedings of the 2017 IEEE conference on visual analytics science and technology. DOI: 10.1109/VAST.2017.8585721.1710.10777.
31.
SpinnerTSchlegelUSchaferH, et al. ExplAIner: a visual analytics framework for interactive and explainable machine learning. IEEE Trans Visual Comput Graph2020; 26: 1064–1074.
32.
RibeiroMTSinghSGuestrinC.“Why should I trust you?” Explaining the predictions of any classifier. In: Proc. ACM SIGKDD Int. conf. on knowledge discovery and data mining. DOI: 10.1145/2939672.2939778.
33.
LundbergSMLeeSI.A unified approach to interpreting model predictions. In: Advances in neural information processing systems. 2017, p.30.
34.
MingYQuHBertiniE.RuleMatrix: visualizing and understanding classifiers with rules. IEEE Trans Vis Comput Graph2019; 25(1): 342–352. DOI: 10.1109/TVCG.2018.2864812
35.
ZhangJWangYMolinoP, et al. Manifold: a model-agnostic framework for interpretation and diagnosis of machine learning models. IEEE Trans Vis Comput Graph2019; 25: 364–373.
36.
KwonBCEysenbachBVermaJ, et al. Clustervision: visual supervision of unsupervised clustering. IEEE Trans Vis Comput Graph2018; 24: 142–151.
37.
JaccardJWanCKTurrisiR.The detection and interpretation of interaction effects between continuous variables in multiple regression. Multivariate Behav Res1990; 25: 467–478.
JaccardJ.Interaction effects in logistic regression. Quantitative Applications in the Social Sciences, Sage Publications, 2001. DOI: 10.4135/9781412984515
40.
VanderWeeleTJKnolMJ.A tutorial on interaction. Epidemiol Methods2014; 3(1): 33–72. DOI: 10.1515/em-2013-0005
41.
FriedmanJHPopescuBE.Predictive learning via rule ensembles. Ann Appl Stat2008; 2(3): 916–954. DOI: 10.1214/07-aoas148
BrittonM.VINE: visualizing statistical interactions in black box models. arXiv preprint arXiv:190400561, 2019. DOI: 10.48550/arXiv.1904.00561.1904.00561.
44.
GoldsteinAKapelnerABleichJ, et al. Peeking inside the Black Box: visualizing statistical learning with plots of individual conditional expectation. J Comput Graph Stat2015; 24(1): 44–65. DOI: 10.1080/10618600.2014.907095.1309.6392
45.
LaminaCSturmGKolleritsB, et al. Visualizing interaction effects: a proposal for presentation and interpretation. Clin Epidemiol2012; 65(8): 855–862.
46.
HookerG.Discovering additive structure in black box functions. In: Proc. ACM SIGKDD int. conference on knowledge discovery and data mining, pp.575–580. DOI: 10.1145/1014052.1014122.
47.
AmerMDaimTUJetterA.A review of scenario planning. Futures2013; 46: 23–40.
48.
HarriesC.Correspondence to what? Coherence to what? What is good scenario-based decision making?Technol Forecast Soc Change2003; 70: 797–817.
49.
GolfarelliMRizziSProliA.Designing what-if analysis: towards a methodology. In DOLAP: Proc. ACM international workshop on data warehousing and OLAP. DOI:10.1145/1183512.1183523.
50.
BarnesCDQuiasonJLBensonC, et al. Success stories in simulation in health care. In: Winter simulation conference proceedings. DOI: 10.1145/268437.268772.
51.
BohanecMZupanBRajkovičV.Applications of qualitative multi-attribute decision models in health care. Int J Med Inform2000; 58-59: 191–205.
SwartRJRaskinPRobinsonJ.The problem of the future: Sustainability science and scenario analysis. Glob Environ Change2004; 14(2): 137–146.
54.
ReillyMWillenbockelD.Managing uncertainty: a review of food system scenario analysis and modelling. Philos Trans Royal Soc B Biol Sci2010; 365(1554): 3049–3063.
55.
van SluisveldMAEHofAFCarraraS, et al. Aligning integrated assessment modelling with socio-technical transition insights: an application to low-carbon energy scenario analysis in Europe. Technol Forecast Soc Change2020; 151: 119177.
56.
SalasEPriestHAWilsonKA, et al. Scenario-based training: improving military mission performance and adaptability. In: AdlerABCastroCABrittTW (eds.) Military life: the psychology of serving in peace and combat (Vol. 2): operational stress. Praeger Security International. Westport, CT: Praeger Security International, 2006, pp.32–53.
57.
KarvetskiCWLambertJHLinkovI.Scenario and multiple criteria decision analysis for energy and environmental security of military and industrial installations. Integr Environ Assess Manag2011; 7(2): 228–236.
58.
HussWRHontonEJ.Scenario planning—what style should you use?Long Range Plann1987; 20(4): 21–29.
59.
HussWR.A move toward scenario analysis. Int J Forecast1988; 4(3): 377–388.
60.
KrauseJPererANgK.Interacting with predictions: visual inspection of black-box machine learning models. In: Proceedings of the CHI conference on human factors in computing systems, 2016, pp. 5686–5697. DOI: 10.1145/2858036.2858529.
61.
WexlerJPushkarnaMBolukbasiT, et al. The what-if tool: interactive probing of machine learning models. IEEE Trans Vis Comput Graph2020; 26: 56–65.
EdwardsW.The theory of decision making. Psychol Bull1954; 51(4): 380–417.
65.
SlovicPLichtensteinSFischhoffB.Decision making. In: AtkinsonRCHerrnsteinRJLindzeyG, et al. (eds.) Stevens’ handbook of experimental psychology. New York: Wiley, 1988, pp.673–738, 2nd ed.Vol. 2, http://hdl.handle.net/1794/22321
66.
KahnemanDTverskyA.Prospect theory: an analysis of decision under risk. In: MacLeanLCZiembaWT (eds.) Handbook of the fundamentals of financial decision making: Part I. Hackensack: World Scientific, 2013, pp.99–127.
67.
TverskyAKahnemanD.The framing of decisions and the psychology of choice. In: WrightG (ed.) Behavioral decision making. Boston, MA: Springer, 1985, pp.25–41.
68.
TverskyAKahnemanD.Rational choice and the framing of decisions. In: KarpakBZiontsS (eds) Multiple criteria decision making and risk analysis using microcomputers. Berlin; New York: Springer, 1989, pp.81–126.
69.
KahnemanDSlovicSPSlovicP, et al. Judgment under uncertainty: Heuristics and biases. Cambridge: Cambridge University Press, 1982.
70.
MilkmanKLChughDBazermanMH.How can decision making be improved?Perspect Psychol Sci2009; 4(4): 379–383. x.
71.
WilsonHDaughertyP.Collaborative intelligence: Humans and AI are joining forces. Harv Bus Rev2018; 96(4): 114–123.
72.
BastaniHBastaniOSinchaisriWP.Improving human decision-making with machine learning. arXiv preprint arXiv:210808454, 2021. DOI: 10.48550/arXiv.2108.08454.
73.
KhosraviKShahabiHPhamBT, et al. A comparative assessment of flood susceptibility modeling using multi-criteria decision-making analysis and machine learning methods. J Hydrol2019; 573: 311–323.
74.
SecinaroSCalandraDSecinaroA, et al. The role of artificial intelligence in healthcare: a structured literature review. BMC Med Inform Decis Mak2021; 21(1): 125–223.
75.
VeropoulosK.Machine learning approaches to medical decision making. PhD Thesis, University of Bristol, England, 2001.
76.
SahooAKPradhanCDasH.Performance evaluation of different machine learning methods and deep-learning based convolutional neural network for health decision making. In: Nature inspired computing for data science. Springer, 2020, pp.201–212.
77.
ThirumalaiCDubaAReddyR.Decision making system using machine learning and pearson for heart attack. In: International conference on electronics, communication and aerospace technology, 2017, pp.206–210, vol. 2. New York, NY: IEEE. DOI: 10.1109/ICECA.2017.8212797.
78.
DuanYEdwardsJSDwivediYK.Artificial intelligence for decision making in the era of Big Data – evolution, challenges and research agenda. Int J Inf Manag2019; 48: 63–71.
79.
DwivediYKHughesLIsmagilovaE, et al. Artificial intelligence (ai): Multidisciplinary perspectives on emerging challenges, opportunities, and agenda for research, practice and policy. Inf Manage2021; 57: 101994.
80.
SimonSMittelst¨adtSKeimDA, et al. Bridging the gap of domain and visualization experts with a liaison. In: Eurographics conference on visualization (EuroVis) short papers. The Eurographics Association. DOI:10.2312/eurovisshort.20151137.
81.
MunznerT.A nested model for visualization design and validation. IEEE Trans Vis Comput Graph2009; 15: 921–928.
82.
AboutorabiHMRKungL. Application of coupled structural acoustic analysis and sensitivity calculations to a tire noise problem. Tire Sci Technol2012; 40(1): 25–41.
83.
LuYGarciaRHansenB, et al. The state-of-the-Art in predictive visual analytics. Comput Graph Forum2017; 36(3): 539–562.
84.
SachaDKrausMKeimDA, et al. VIS4ML: an ontology for visual analytics assisted machine learning. IEEE Trans Vis Comput Graph2019; 25(1): 385–395. DOI: 10.1109/TVCG.2018.2864838
85.
ZhouJArshadSZLuoS, et al. Effects of uncertainty and cognitive load on user trust in predictive decision making. In: Human-computer interaction–INTERACT 2017: 16th IFIP TC 13 international conference, Mumbai, India, 25–29 September 2017, pp.23–39, Proceedings, Part IV 16. Springer.
86.
KindtPSasPDesmetW.Measurement and analysis of rolling tire vibrations. Opt Lasers Eng2009; 47(3-4): 443–453.
87.
WillettWHeerJAgrawalaM.Scented widgets: improving navigation cues with embedded visualizations. IEEE Trans Vis Comput Graph2007; 13(6): 1129–1136.
NewmanGESchollBJ.Bar graphs depicting averages are perceptually misinterpreted: the within-the-bar bias. Psychon Bull Rev2012; 19(4): 601–607.
91.
CoppersSLuytenKVanackenD, et al. Fortunettes: feedforward about the future state of GUI widgets. Proc ACM Hum-Comput Interact2019; 3: 1–20.
92.
Gutierrez-GomezLPetryFKhadraouiD.A comparison framework of machine learning algorithms for mixed-type variables datasets: a case study on tire-performances prediction. IEEE Access2020; 8: 214902–214914.
93.
GleicherMAlbersDWalkerR, et al. Visual comparison for information visualization. Inf Vis2011; 10(4): 289–309.
94.
PiñeiroGPerelmanSGuerschmanJP, et al. How to evaluate models: observed vs. Predicted or predicted vs. observed?Ecol Modell2008; 216: 316–322.
95.
CiesielskiK.Set theory for the working mathematician. Cambridge: London Mathematical Soc. Student Texts, Cambridge University Press, 1997.
96.
du PrelJBHommelGRöhrigB, et al. Confidence interval or p-value?: part 4 of a series on evaluation of scientific publications. Dtsch Arztebl Int2009; 106: 335–339.
97.
HarterHL.The method of least squares and some alternatives: Part V. Int Stat Rev1975; 43: 269.
98.
FoxJWeisbergS.Fitting linear models. JMP, A Business Unit of SAS2011.
99.
GhoniemMFeketeJDCastagliolaP.A comparison of the readability of graphs using node-link and matrix-based representations. In: Proceedings of the IEEE symposium on information visusalization, INFOVIS. DOI: 10.1109/INFVIS. 2004.1.
100.
BoosDDStefanskiLA.P-value precision and reproducibility. Am Stat2011; 65: 213–221.
101.
HuberPJ.Projection pursuit. Ann Stat2007; 13(2): 435–475.
102.
PearsonK.Notes on the history of correlation. Biometrika1920; 13: 25–45.
103.
SiramdasuYLiKWheelerR.Understanding tire dynamic characteristics for vehicle dynamics ride using simulation methods. Tire Sci Technol2020; 48(3): 188–206.
104.
MerkelD.Docker: lightweight linux containers for consistent development and deployment. Linux J2014; 2014(239): 2.
105.
LamHBertiniEIsenbergP, et al. Empirical studies in information visualization: seven scenarios. IEEE Trans Vis Comput Graph2012; 18(9): 1520–1536.
106.
SaraiyaPNorthCDucaK.An insight-based methodology for evaluating bioinformatics visualizations. IEEE Trans Vis Comput Graph2005; 11(4): 443–456.
107.
StaskoJ.Value-driven evaluation of visualizations. In: Proceedings of the fifth workshop on beyond time and errors: novel evaluation methods for visualization, BELIV ’14, ACM, 2014, pp.46–53. DOI: 10.1145/2669557.2669579.
108.
IsenbergTIsenbergPChenJ, et al. A systematic review on the practice of evaluating visualization. IEEE Trans Vis Comput Graph2013; 19(12): 2818–2827.
109.
ElmqvistNYiJS.Patterns for visualization evaluation. Inf Vis2015; 14(3): 250–269.
110.
CarpendaleS.Evaluating information visualizations. In: KerrenAStaskoJTFeketeJD, et al. (eds) Information Visualization. Lecture Notes in Computer Science. Berlin, Heidelberg: Springer, 2008, pp.19–45, Vol. 4950.
111.
WeberGHCarpendaleSEbertD, et al. Apply or die: on the role and assessment of application papers in visualization. IEEE Comput Graph Appl2017; 37(3): 96–104.
112.
MazzaRBerreA.Focus group methodology for evaluating information visualization techniques and tools. In: Proceedings of the international conference on information visualization (IV’07), 2007, pp.74–80. New York, NY: IEEE. DOI: 10.1109/IV.2007.51.
TaylorJ.Classical mechanics. G-Reference, information and interdisciplinary subjects series. University Science Books, 2005.
122.
NagayaKIkaiSChibaM, et al. Tire with self-repairing mechanism. JSME Int J Ser C Mech Syst Mach Element Manuf2006; 49(2): 379–384.
123.
Student. The probable error of a mean. Biometrika1908; 6: 1–25.
124.
De WinterJ. Using the student’s t-test with extremely small sample sizes. Pr Assess Res Eval2013; 18(1): 10.
125.
TavazzaFDeCostBChoudharyK.Uncertainty prediction for machine learning models of material properties. ACS Omega2021; 6(48): 32431–32440.
126.
GuyonI, et al. Practical feature selection: from correlation to causality. In: Mining massive data sets for security: advances in data mining, search, social networks and text mining, and their applications to security. 2008, pp.27–43. DOI: 10.3233/978-1-58603-898-4-27
127.
De GonzalezABCoxDR. Interpretation of interaction: a review. Ann Appl Stat2007; 1(2): 371–385.
128.
ChakhchoukhMBoukhelifaNBezerianosA.Understanding how in-visualization provenance can support trade-off analysis. IEEE Trans Vis Comput Graph2023; 29: 3758–3774.
129.
GotzDZhouMX.Characterizing users’ visual analytic activity for insight provenance. In: 2008 IEEE symposium on visual analytics science and technology, 2008, pp.123–130. New York, NY: IEEE.
130.
SegelEHeerJ.Narrative visualization: telling stories with data. IEEE Trans Vis Comput Graph2010; 16(6): 1139–1148.
131.
StijvenSMinneboWVladislavlevaK.Separating the wheat from the chaff: on feature selection and feature importance in regression random forests and symbolic regression. In: Proceedings of the annual conference companion genetic evolutionary computing, 2011, pp.623–630. DOI: 10.1145/2001858.2002059.
132.
SedlmairMMeyerMMunznerT.Design study methodology: reflections from the trenches and the stacks. IEEE Trans Vis Comput Graph2012; 18(12): 2431–2440.
133.
PajerSStreitMTorsney-WeirT, et al. Weightlifter: visual weight space exploration for multi-criteria decision making. IEEE Trans Vis Comput Graph2017; 23(1): 611–620.
134.
LaurillauYNguyenVBCoutazJ, et al. The TOP-slider for multi-criteria decision making by non-specialists. In: ACM Int. Conf. Proceeding Series, NordiCHI ’18, 2018, pp.642–653. ACM. DOI: 10.1145/3240167.3240185.
135.
NadareishviliIMitraRMcLartyM, et al. Microservice architecture: aligning principles, practices, and culture. O’Reilly Media, Inc, 2016.
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
