Dwight Dinkins

I use my expertise in Mathematics and Mathematical Modeling to improve contemporary Machine Learning Models for better prediction and decision-making. Mathematical Models is an instrument for improving the lives of ordinary people, which is my incentive. I have a wealth of knowledge devoted to innovation and mathematical modeling.

Title: Entry-Level Applied Scientist

• National Honors Society
• Excellence in Mathematics

Does Regular Exercise Reduce Stress?:Developed a RandomForestRegressor, using data collected from FitBit Versa 4,  to predict stress levels. Found lack of sleep to be the main contributor.

California Housing Dataset: While at RIT, I enrolled in Foundations of Data Science, in which I was assigned the California Housing Dataset as the course final project. I chose RandomForestRegressor and MLPRegressor, as the dataset consisted of nonlinear features: Latitude and Longitude. I obtained a mean-squared-error (MSE) of 0.1 on the training dataset using Sci-kit Learn's MLPRegressor and an MSE of 0.2 on the testing dataset. 

Second-Order Permutation Variable Importances for Random Forest Regressor: My thesis paper at RIT pertains to the quantification and attribution of uncertainty of future coastal risk. I found a relationship between Sobol Sensitivity Indices (or measures) and Second-Order Permutation variable Importances for RandomForestRegressor (RFR), overcoming the bias present within RFR's feature importance. 

Mathematical Model for Substituting Insect Frass for Mineral Fertilizer:  Demonstrated that insect frass isn’t a viable alternative for mineral  fertilization, as it requires large buildings (bigger than the Ei el Tower) to  house them and to produce enough frass for a farm with 2400 acres of land. 

Modeling the effects of decomposing cow carcasses (leachate) on groundwater basins in Tooele, Utah: Developed Mathematical Model, adapting Darcy’s Law of Flow. Found that the amount of cow-carcass contaminants(leachate) in the groundwater will be  9.46 ppm(parts per million), which exceeded the USEPA’s recommendations. Recommended the installation of filters.