In an effort to address the lack of literature in on-campus active travel, this study aims to investigate the following primary questions:<br/>• What are the modes that students use to travel on campus?<br/>• What are the motivations that underlie the mode choice of students on campus?<br/>My first stage of research involved a series of qualitative investigations. I held one-on-one virtual interviews with students in which I asked them questions about the mode they use and why they feel that their chosen mode works best for them. These interviews served two functions. First, they provided me with insight into the various motivations underlying student mode choice. Second, they provided me with an indication of what explanatory variables should be included in a model of mode choice on campus.<br/>The first half of the research project informed a quantitative survey that was released via the Honors Digest to attract student respondents. Data was gathered on travel behavior as well as relevant explanatory variables.<br/>My analysis involved developing a logit model to predict student mode choice on campus and presenting the model estimation in conjunction with a discussion of student travel motivations based on the qualitative interviews. I use this information to make a recommendation on how campus infrastructure could be modified to better support the needs of the student population.
Currently, autonomous vehicles are being evaluated by how well they interact with humans without evaluating how well humans interact with them. Since people are not going to unanimously switch over to using autonomous vehicles, attention must be given to how well these new vehicles signal intent to human drivers from the driver’s point of view. Ineffective communication will lead to unnecessary discomfort among drivers caused by an underlying uncertainty about what an autonomous vehicle is or isn’t about to do. Recent studies suggest that humans tend to fixate on areas of higher uncertainty so scenarios that have a higher number of vehicle fixations can be reasoned to be more uncertain. We provide a framework for measuring human uncertainty and use the framework to measure the effect of empathetic vs non-empathetic agents. We used a simulated driving environment to create recorded scenarios and manipulate the autonomous vehicle to include either an empathetic or non-empathetic agent. The driving interaction is composed of two vehicles approaching an uncontrolled intersection. These scenarios were played to twelve participants while their gaze was recorded to track what the participants were fixating on. The overall intent was to provide an analytical framework as a tool for evaluating autonomous driving features; and in this case, we choose to evaluate how effective it was for vehicles to have empathetic behaviors included in the autonomous vehicle decision making. A t-test analysis of the gaze indicated that empathy did not in fact reduce uncertainty although additional testing of this hypothesis will be needed due to the small sample size.
Grading schemes for breast cancer diagnosis are predominantly based on pathologists' qualitative assessment of altered nuclear structure from 2D brightfield microscopy images. However, cells are three-dimensional (3D) objects with features that are inherently 3D and thus poorly characterized in 2D. Our goal is to quantitatively characterize nuclear structure in 3D, assess its variation with malignancy, and investigate whether such variation correlates with standard nuclear grading criteria.
Methodology
We applied micro-optical computed tomographic imaging and automated 3D nuclear morphometry to quantify and compare morphological variations between human cell lines derived from normal, benign fibrocystic or malignant breast epithelium. To reproduce the appearance and contrast in clinical cytopathology images, we stained cells with hematoxylin and eosin and obtained 3D images of 150 individual stained cells of each cell type at sub-micron, isotropic resolution. Applying volumetric image analyses, we computed 42 3D morphological and textural descriptors of cellular and nuclear structure.
Principal Findings
We observed four distinct nuclear shape categories, the predominant being a mushroom cap shape. Cell and nuclear volumes increased from normal to fibrocystic to metastatic type, but there was little difference in the volume ratio of nucleus to cytoplasm (N/C ratio) between the lines. Abnormal cell nuclei had more nucleoli, markedly higher density and clumpier chromatin organization compared to normal. Nuclei of non-tumorigenic, fibrocystic cells exhibited larger textural variations than metastatic cell nuclei. At p<0.0025 by ANOVA and Kruskal-Wallis tests, 90% of our computed descriptors statistically differentiated control from abnormal cell populations, but only 69% of these features statistically differentiated the fibrocystic from the metastatic cell populations.
Conclusions
Our results provide a new perspective on nuclear structure variations associated with malignancy and point to the value of automated quantitative 3D nuclear morphometry as an objective tool to enable development of sensitive and specific nuclear grade classification in breast cancer diagnosis.
Throughout this project, I decided on a number of learning goals to consider it a success. I needed to learn how to use the supporting libraries that would help me to design this system. I also learned how to use the Twitter API, as well as create the infrastructure behind it that would allow me to collect large amounts of data for machine learning. I needed to become familiar with common machine learning libraries in Python in order to create the necessary algorithms and pipelines to make predictions based on Twitter data.
This paper details the steps and decisions needed to determine how to collect this data and apply it to machine learning algorithms. I determined how to create labelled data using pre-existing Botometer ratings, and the levels of confidence I needed to label data for training. I use the scikit-learn library to create these algorithms to best detect these bots. I used a number of pre-processing routines to refine the classifiers’ precision, including natural language processing and data analysis techniques. I eventually move to remotely-hosted versions of the system on Amazon web instances to collect larger amounts of data and train more advanced classifiers. This leads to the details of my final implementation of a user-facing server, hosted on AWS and interfacing over Gmail’s IMAP server.
The current and future development of this system is laid out. This includes more advanced classifiers, better data analysis, conversions to third party Twitter data collection systems, and user features. I detail what it is I have learned from this exercise, and what it is I hope to continue working on.