Esther's Project Page

Who:  I am originally from Arizona but have been living in Utah for the last 5 years. I chose to attend University of Utah for the experience with ACCESS as well as the amazing research and engineering opportunities offered here. I also love traveling, learning new languages, and boxing competitively.

My scientific/engineering interests:  I’ve known that I wanted to study engineering ever since I was a freshman in high school, and was able to tour NASA’s Jet Propulsion Laboratory (JPL). I knew immediately that I wanted to help create things that could help people and inspire others the way that I had been inspired.

Academic goals:  I am now majoring in mechanical engineering. I hope that I develop as a coder and researcher, and become published over the course of my undergraduate career. After I graduate, I intend to pursue a PhD in a related field, ideally abroad.

Career goals:  Something else that I’ve been considering for a long time, is the idea of pursuing a career in higher education and research or focusing more on a career in industry. Often I ponder on what will fit me best, but so far all I can say is that I am passionate about my major and the best approach is just to gain as much experience as possible.

Machine learning is the study of algorithms that detect patterns in large sets of data and are able to make statistical predictions for future data. By using already established temperature-dependent heat capacity data and processing it through a Random Forest Regressor algorithm (which creates a decision tree model based on the training data’s features), the algorithm is able to predict the heat capacities of the remaining testing data based on the features of the training data. This is a process called data splitting which works by dividing the initial dataset into training data and test data; the training data is used by the algorithm to featurize the data and learn statistically significant features while the test data remains unseen until the final evaluation. Immediately following is the model training process which focuses on cleaning the data and establishing hyperparameters that are meant to safeguard against overfitting the data. In this project, we focus mainly on modeling the predicted heat capacities of the test data against the true heat capacities of the test data to test the accuracy of the RFR algorithm’s predictions. Up until recently, temperature-dependent heat capacity was largely being estimated through processes that require specialized equipment and laborious experimentation, which has resulted in an insufficient range of data. The goal of this research is to show how machine learning can be a tool in the prediction of material properties, such as the temperature-dependent heat capacities, in the development of new materials.

Deploying Machine Learning Model for the Rapid Prediction of Heat Capacity Download Deploying Machine Learning Model for the Rapid Prediction of Heat Capacity