Yinzhen L.

Key Skills

Software

No software listed

Top Subject Matter

Nlp Domain Expertise

social media sentiment/topic analysis

Clinical prediction

Top Data Types

Text

Top Task Types

Emotion Recognition

Diagnosis

Mapping

Classification

Data Collection

Freelancer Overview

Adoption of Artificial Intelligence in Online Communities: A Socio-Technical Perspective (Researcher). Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include snowNLP, BERT, and LDA. Education includes Bachelor of Science, Beijing Jiaotong University (BJTU) (2026) and Bachelor of Science, Rochester Institute of Technology (RIT) (2026). AI-training focus includes data types such as Text, Medical, and DICOM and labeling workflows including Emotion Recognition, Diagnosis, and Mapping.

Labeling Experience

Unemployment Rate Prediction Model Based on Interpretable Machine Learning (Project Initiator)

TextData Collection

Initiated a project that gathered macroeconomic data into a modeling-readable structure suitable for interpretability-focused analysis. Addressed data quality issues by applying multiple imputation techniques to handle missing values. Built and compared predictive models and then conducted an interpretability study on the best-performing model. • Collected macroeconomic data from open sources such as CEIC • Formatted raw data into model-readable structures • Applied multiple imputation for missing values • Evaluated models (R², RMSE, MAE) and performed interpretability study on the top model

2024 - Present

A Typhoon Classifier Model Based on Interpretable Machine Learning and K-means (Project Initiator)

Classification

Extracted and processed NOAA data, then performed feature selection to prepare inputs for an interpretable typhoon classification workflow. Encoded the geospatial dataset into GeoJSON format using Geopy to support consistent model ingestion. Trained multiple classifier models and interpreted the chosen model using SHAP for explainable outputs. • NOAA data extraction and feature selection • GeoJSON encoding using Geopy • Trained classifiers including ANN, SVM, KNN, CatBoost, RF, LR, and XGBoost • SHAP-based interpretation of the selected model

2024 - 2025

An Interpretable Typhoon Trajectory Prediction Model Based on High-Dimensional Time-series Geo-XGBoost (Project Initiator)

Mapping

Extracted typhoon-related data from the NOAA database and processed it with feature selection for trajectory prediction. Encoded the data into GeoJSON format using Geopy to standardize geospatial inputs for modeling. Compared models to select the optimal predictor and interpreted feature contributions using SHAP. • NOAA data extraction and preprocessing • Feature selection for high-dimensional time-series modeling • GeoJSON encoding via Geopy • SHAP-based interpretation of the selected model

2024 - 2025

Adoption of Artificial Intelligence in Online Communities: A Socio-Technical Perspective (Researcher)

TextEmotion Recognition

Spearheaded extraction of user comments about AI platforms from major Chinese social media sites including Weibo, CSDN, and Zhihu. Built labeling and analytic targets to support sentiment and topic-based downstream evaluation. Used this prepared dataset to drive interpretable socio-technical findings in an academic paper. • Data collection from public social media comment text • Applied sentiment analysis with snowNLP and lexicon-based emotion dictionaries • Conducted topic-based analysis using LDA and BERT models • Authored methodology and results for the paper describing the processing pipeline

2024 - 2024

A New Evaluation Model for Traumatic Severe Pneumothorax Based on Interpretable Machine Learning (Principal Researcher)

Diagnosis

Extracted and preprocessed patient data from the MIMIC-IV and eICU databases using explicit inclusion and exclusion criteria. Performed data cleaning and missing-value handling through multiple imputation to make the dataset analysis-ready. Developed and evaluated predictive models and emphasized interpretability of the best-performing approach for clinical applicability. • Patient data extraction with defined inclusion/exclusion criteria • Multiple imputation to address missing values • Model development and fine-tuning with performance assessment • Authored the majority of the paper focusing on model interpretability

2023 - 2024

Education

Work History