Statistics | Regression | Machine learning

Data projects with the evaluation built in.

I am Lei Duli, a statistics student at the University of Arizona. I build reproducible analysis projects that connect modeling choices, diagnostics, and real-world interpretation.

View projects Contact me

48,895

Airbnb listings analyzed

0.507

best adjusted R2

0.922

best ROC AUC

41.9%

high-intent conversion rate

Featured work

Selected Projects

Course projects turned into portfolio pieces: clear questions, reproducible code, model comparison, diagnostics, and reports.

Boxplots of Airbnb log price by room type and borough

DATA 467 | Linear Regression

Predicting Airbnb Listing Prices in New York City

A hedonic price modeling project using 2019 NYC Airbnb listings to explain nightly price through room type, borough, review activity, host scale, and availability.

38,833 cleaned listings 0.507 best adj. R2 118% entire-home premium

Python OLS GLM Diagnostics Hedonic pricing

Project page Notebook Final paper Proposal

Model performance chart for online purchase intention project

STAT 474 | Modeling Strategy and Evaluation

From Browsing to Buying: Predicting Online Purchase Intention

A machine learning classification project using e-commerce session behavior to predict purchase intent under strong class imbalance.

6 model families 0.667 best F1 0.922 best ROC AUC

R caret Random Forest Gradient Boosting F1 thresholding

Project page R code Final report Proposal

Data preprocessing

Large-Scale Data Preprocessing and Feature Discretization

Prepared a used-car transaction dataset through missing value handling, outlier detection, discretization, and encoding.

Python pandas scikit-learn Feature engineering

Research synthesis

Heuristic Optimization of LSTM-Based Models

Reviewed PSO, GA, SA, and ACO optimization strategies for LSTM-based forecasting across applied time-series domains.

LSTM PSO GA Time series

Technical toolkit

What I work with

Languages

R, Python, SQL

Data work

EDA, cleaning, feature engineering, visualization, reporting

Modeling

OLS, GLM, logistic regression, LASSO, PCA, random forest, gradient boosting, SVM

Evaluation

Cross-validation, diagnostics, ROC AUC, F1, precision, recall, balanced accuracy