Selena Ding Email & Phone Number

New York, Ny, Us

New York, United States

Cysyphus: Cybersecurity policy recommendation tool

New York, United States

Worked in a multidisciplinary team analyzing the impact of sex education policies on various social and health outcomes with confounding factors across each state in the U.S.Applied a combination of embedding and feature extraction techniques, including FastText, word2vec, bag of words, SentenceBERT, BERT (for embedding), and NMF, TF-IDF, LDA (for feature extraction) in text analysis. Utilized these methods to calculate cosine similarity scores, identifying similar pairs in policy texts.

New York, United States

Conducted a detailed study quantifying the critical role of women in biomass production across key developing regions: Sub-Saharan Africa, Developing Asia, and Latin America underscoring their vital role in energy production.Meticulously sourced, vetted, and collated extensive datasets from reputable sources including the International Energy Agency (IEA), the World Bank, and the Global Data Lab. Employed rigorous data handling techniques to ensure accuracy and reliability, assembling a… Show more Conducted a detailed study quantifying the critical role of women in biomass production across key developing regions: Sub-Saharan Africa, Developing Asia, and Latin America underscoring their vital role in energy production.Meticulously sourced, vetted, and collated extensive datasets from reputable sources including the International Energy Agency (IEA), the World Bank, and the Global Data Lab. Employed rigorous data handling techniques to ensure accuracy and reliability, assembling a robust dataset for analysis.Utilized advanced functionalities in Excel for comprehensive data analysis, focusing on identifying populations and households lacking access to clean cooking facilities, and highlighting critical areas for intervention and support. Show less

Davis, California, United States

• Visualized the daily vaccination rate for each state in the United States by splitting cumulative vaccination rate data and plotted a graph for each state to find outliers and remove them. Collected data on the percentage of votes for Democrat candidates in the US Presidential Election 2020 from the United States Census Bureau.• Employed FPCA to the daily vaccination rate and found the first three FPCs could explain 98% of the variance and use only these three FPCs in the regression… Show more • Visualized the daily vaccination rate for each state in the United States by splitting cumulative vaccination rate data and plotted a graph for each state to find outliers and remove them. Collected data on the percentage of votes for Democrat candidates in the US Presidential Election 2020 from the United States Census Bureau.• Employed FPCA to the daily vaccination rate and found the first three FPCs could explain 98% of the variance and use only these three FPCs in the regression part.• Built regression between the percentage of votes for Democrat candidates in the US Presidential Election 2020 and each FPC and found only the first two FPCs are valid in regression by looking at the p-value in regression.• Implemented z from 20% to 70% for the predictor to see how the daily vaccination rate would change and found as the percentage of votes for Democrat candidates in the US Presidential Election 2020 increases, the vaccination is taken earlier, and for most times, the higher the percentage of votes for Democrat candidate in the US Presidential Election 2020, the higher the daily vaccination rate is. Show less

University Of California Davis, California, United States

• Researched and aggregated time series datasets by R programming, each dataset includes 1,577,400+ data on Covid-19 confirmed cases and death cases of each state in the United States from November 2020 to February 2021, and datasets on population, area, google mobility, and GDP per capita for each state. Worked on data cleaning and calculated 52 states' population density.• Split cumulative covid dataset to obtain a clear outline of each day's newly confirmed cases, and death cases. Employ… Show more • Researched and aggregated time series datasets by R programming, each dataset includes 1,577,400+ data on Covid-19 confirmed cases and death cases of each state in the United States from November 2020 to February 2021, and datasets on population, area, google mobility, and GDP per capita for each state. Worked on data cleaning and calculated 52 states' population density.• Split cumulative covid dataset to obtain a clear outline of each day's newly confirmed cases, and death cases. Employ R to visualize over 4000 observations of the state's COVID record and draw plots to observe cases' outbreaking peaks.• Calculated density function and quantile function of COVID record for 52 states and designed functions to smooth the density function and remove boundary effect.• Applied R to do optimal transport from over 4000 observations of confirmed cases to death cases and deploy the GloDenReg package in R to do regression on multiple factors.• Developed regression models on optimal transport densities and 4 covariates, including population density, GDP per capita, and google mobility, and identified regression relationships. Show less

University Of California Davis, California, United States

- Collected 75,000+ economic data points by leveraging google spreadsheets.- Researched and examined over 1,000 potential reviewers' CVs, documented corresponding Ph.D. institutions and employment, and aided Professor Lester to do data cleaning.

Seattle, Washington, United States

• Visualized data about customer service interactions, including customer information and whether the customer’s problem is solved in the service by Python.• Examined missing values in the data and properly dealt with them by dropping or replacing them with the attribute means, most frequent (for categorical) or other values decided by case. Encoded the nominal categorical data by one-hot encoding.• Designed a decision tree model to predict whether a problem will be solved in the… Show more • Visualized data about customer service interactions, including customer information and whether the customer’s problem is solved in the service by Python.• Examined missing values in the data and properly dealt with them by dropping or replacing them with the attribute means, most frequent (for categorical) or other values decided by case. Encoded the nominal categorical data by one-hot encoding.• Designed a decision tree model to predict whether a problem will be solved in the customer service for the training data setwith a total of 5800+ entries and ran validation on the model. Evaluated the model by F1 score.• Standardized encoded training data and developed a Logistic regression model on the training set of over 1000 observations by the sklearn package. Performed k-cross-validation and evaluated the model by F1 score.• Tuned parameters of the models and got a final F1 score of 62.58% while the highest reachable F1 score is around 70%. Show less

Master Of Science - Ms, Data Science

Bachelor'S Degree, Applied Mathematics, 3.93/4.0

Activities and Societies: Honors: Honors at graduation in Statistics, Honors at graduation in Applied Mathematics, Dean's Honors List.

Summer Session, Mathematics, 4.0/4.0