Daniil Azarov

California Housing Analysis

Developed a predictive framework for California housing prices using the 1990 California Housing dataset. Implemented data cleaning, feature engineering (including distance and elite area features), exploratory data analysis, and visualization techniques to analyze housing patterns. Compared regression methods including linear models, Ridge, Lasso, Random Forest, and XGBoost, with model evaluation based on R² scores and feature importance. Technologies used: Python, pandas, numpy, scikit-learn, XGBoost, matplotlib, seaborn, SciPy.

F1 Drivers Analysis

Developed a comparison framework for Formula 1 drivers' ability to finish a race. Implemented data cleaning, feature engineering, exploratory data analysis, and visualization techniques to analyze driver statistics. Methods included logistic regression and ensemble methods (random forests, boosting - XGBoost), and model evaluation using ROC-AUC and confusion matrices. Technologies used: Python, pandas, numpy, scikit-learn, matplotlib, seaborn.

Loan Default Predictions

Built a comprehensive data analysis to predict loan default risk using a dataset of loan applications. The project involved data preprocessing, feature engineering, exploratory data analysis, feature selection, and predictive modeling using logistic regression, linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), naive Bayes classifier, and ensemble methods (random forests, gradient boosting). Evaluated model performance with accuracy, precision, recall, and ROC-AUC metrics. Created visualizations to communicate findings effectively. Technologies used: R, dplyr, ggplot2, car, MASS, randomForest, gbm, pROC.

Visual Memorability Research

Surprisingly, despite individual differences, our visual memory is very consistent across people. Different individuals tend to remember the same objects with similar accuracy - if I remember this lamp well, you probably will too. People also show similar false memories: if it seems to me that I saw a chair when I didn't, it will likely seem that way to you as well.
My research explores how context, object similarity, and task conditions affect what we remember.

See: Utochkin, Azarov, Grigorev (2025)

Ensemble Perception

People have an amazing ability to quickly and effortlessly grasp the overall pattern of what they see. For example, we can instantly sense how fast traffic is moving or how fresh the berries at the market look on average. This skill - known as ensemble perception - allows us to extract the “gist” of a group of objects without focusing on each one individually.
My research explores how far we can take this ability: what kinds of information can our brains summarize at a glance, and where are the limits of this remarkable visual skill?

See: Azarov, Sanborn, Chater, Goldstone (2025); Azarov, Grigorev, Utochkin (2024)