Dejun (Jessie) Kong
Contact: 979-***-**** • admeb2@r.postjobfree.com • 5208 Newbury Way, Columbia, MO 65203 Permanent Resident
Competencies
• In-depth knowledge and hands-on experience on Statistical Analysis, Regression, Predictive Modeling, Machine Learning, Deep Learning and Data Visualization
• Proficiency in SAS, R, Python (NumPy, Pandas, Pytorch, Matplotlib, OpenCV), SQL and Microsoft Office
Education
• Master of Arts, Applied Statistics, University of Missouri, Expected graduation date: 05/2021 Core Courses: Data Analysis I - IV, Experimental Design, Intro Probability Theory, Statistical Inference GPA: 3.96/4.00
• Bachelor of Science, Physics, Anqing Normal University, China Professional Experiences
• Substitute teacher, College Station Independent School District, Texas
• Advertising Department, Potato Education Tech LLC o Graphic designer - From no experience to handling all advertising materials (i.e., website, social media, flyers, banner, uniform)
o Data manager - Managed digital data of business contents and customer relations o Event planner - Planned online/offline show-case events for business growth
• Physics teacher, Hefei 168 high school, Anhui, China o Outstanding teacher of the city of Hefei
o Director, 8th grade Physics Curriculum Development Project Experiences
• Image Segmentation: classifying each and every voxel of HP gas MR images of lung - Master thesis
o Converted Rdata to tiff file, wrote a data generator using Python package ‘Pytorch’ to feed our network (U-net model), and visualized the results using ‘cv2’ for edges detection and image blending
o Code running on Google Colab cloud platform
• Data analysis on the usage of two drugs on COVID-19 patients o Compared random effects using generalized linear mixed model o Compared interactions between time and treatment by adding a non-linear term into the mixed model using SAS
• Author classification of books based on objective features of texts o Applied SVM, PCR, K Nearest Neighbors, tree-based model (Random Forests, Stochastic Boosting), CNN and ensemble model
o A high classification accuracy was obtained by standardizing the training and test data
• House prices prediction - Kaggle competition
o Cleaned the data and performed feature engineering (79 explanatory variables) o Applied cross-validation method to evaluate the performances among linear regression, tree-based models, Lasso, and ridge regression using R
• Volcanic eruption prediction - Kaggle competition o Extracted features including missing values using the ‘tsfeature’ R package from the huge time series data
o Selected the most important and relevant features using the ‘caret’ R package o Applied Lasso, Ridge, PCR, Gradient Boosting Machine, Random Forests, 1D CNN, ANN, and ensemble model for predicting the next volcano eruption time