Search for: All records

Presenting a predictive model's performance is a communication bottleneck that threatens collaborations between data scientists and subject matter experts. Accuracy and error metrics alone fail to tell the whole story of a model – its risks, strengths, and limitations – making it difficult for subject matter experts to feel confident in their decision to use a model. As a result, models may fail in unexpected ways or go entirely unused, as subject matter experts disregard poorly presented models in favor of familiar, yet arguably substandard methods. In this paper, we describe an iterative study conducted with both subject matter experts and data scientists to understand the gaps in communication between these two groups. We find that, while the two groups share common goals of understanding the data and predictions of the model, friction can stem from unfamiliar terms, metrics, and visualizations – limiting the transfer of knowledge to SMEs and discouraging clarifying questions being asked during presentations. Based on our findings, we derive a set of communication guidelines that use visualization as a common medium for communicating the strengths and weaknesses of a model. We provide a demonstration of our guidelines in a regression modeling scenario and elicit feedback on their use from subject matter experts. From our demonstration, subject matter experts were more comfortable discussing a model's performance, more aware of the trade-offs for the presented model, and better equipped to assess the model's risks – ultimately informing and contextualizing the model's use beyond text and numbers. 

In Caenorhabditis elegans, many genes involved in the formation of the cuticle are also known to influence body size and shape. We assessed post-embryonic growth of both long and short C. elegans body size mutants from the L1 to L4 stage. We found similar developmental trajectories of N2 and lon-3 animals. By contrast, we observed overall decreases in body length and increases in body width of tested dpy mutants compared to N2, consistent with the Dpy phenotype. We further show that the dynamics of animal shape in the mutant strains are consistent with a previously proposed “Stretcher” growth model. 

ABSTRACT Swimming organisms may actively adjust their behavior in response to the flow around them. Ocean flows are typically turbulent and are therefore characterized by chaotic velocity fluctuations. While some studies have observed planktonic larvae altering their behavior in response to turbulence, it is not always clear whether a plankter is responding to an individual turbulence fluctuation or to the time-averaged flow. To distinguish between these two paradigms, we conducted laboratory experiments with larvae in turbulence. We observed veliger larvae of the gastropod Crepidula fornicata in a jet-stirred turbulence tank while simultaneously measuring two components of the fluid and larval velocity. Larvae were studied at two different stages of development, early and late, and their behavior was analyzed in response to different characteristics of turbulence: acceleration, dissipation and vorticity. Our analysis considered the effects of both the time-averaged flow and the instantaneous flow, around the larvae. Overall, we found that both stages of larvae increased their upward swimming speeds in response to increasing turbulence. However, we found that the early-stage larvae tended to respond to the time-averaged flow, whereas the late-stage larvae tended to respond to the instantaneous flow around them. These observations indicate that larvae can integrate flow information over time and that their behavioral responses to turbulence can depend on both their present and past flow environments.