Skip to content
View lukablagoje's full-sized avatar

Block or report lukablagoje

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
lukablagoje/README.md

Hello, I'm Luka Blagojević! 👋

I'm a Research Scientist, with an educational background in Theoretical and Experimental Physics (BA and MA) and a PhD in Network Science.

In my research, I take on new problems and develop mathematical and algorithmic solutions for systems that have an interconnected structure (e.g. social networks, biological neural networks). Below are my previous academic and applied projects:

Academic Research Projects

The Impact of Physicality on Network Structure

  • Publication: Nature Physics | GitHub
    • As a contributing author, developed collision-detection algorithms for $10^{6}$ objects with approximately 3000 unique labels.
    • Detected spatial neighbors by computing pairwise distances of 3D objects, computed with point clouds and kd-trees.
    • Using graph analysis, found a relationship of neuron synaptic connections to the number of their spatial neighbors.

Temporal Patterns of Reciprocity in Communication Networks

  • Publication: EPJ Data Science | GitHub
    • As a contributing author, cleaned and processed the data of 6 timestamped graph datasets (e.g., emails, Twitter).
    • Performed null hypothesis testing for 4 reference null models (randomized timestamps and graph topology).
    • Discovered statistically significant results that Twitter exchanges are less reciprocal and bursty than SMS, calls, and emails.

Three-Dimensional Shape and Connectivity of Physical Networks

  • Publication: Nature Scientific Reports | GitHub
    • As a leading author, performed a comprehensive data processing and analysis of 15 volumetric 3D graph datasets.
    • Developed algorithms to quantify the shape, size, and geometry of the data (e.g., fractal dimension, edge volume).
    • Created a pipeline that randomizes physical edge trajectories to detect obstacles for more than $10^{5}$ edges at once.

Physical Network Robustness

  • Status: In progress | GitHub
    • As a leading author, simulating 2D and 3D graphs' physical attacks (spatial edge removal) in their embedding space.
    • Developing a measure to quantify the connectivity of spatial regions in which the graph (network) is embedded.

Applied Data Science Projects

Quantifying and Ranking User Engagement with Clickbait Articles Using NLP-Created Features

  • Event: Citadel - Correlation One Global PhD Datathon 2023 | Competition Link | GitHub
    • As an individual competitor, utilized NLP methods to determine sentiment, emotion, and topic of text data.
    • Computed correlations of click-through-rates to determine what drives user engagement, with Google Analytics data.
    • Developed a custom ranking of clickbait articles, that relied on their daily, aggregate, and top 5% performance.

Dynamic XGBoost-Based Model on a Data Stream for Stock Price Prediction

  • Event: Optiver - Trading at the Close (Kaggle) Competition | Kaggle Link | GitHub
    • As a leader of a 3-member team, implemented an XGBoost model on a data stream to predict stock prices.
    • Optimized hyperparameters using k-fold cross-validation tailored for time-series data, including periodic retraining.
    • Automated data-stream tasks: data collection and cleaning, feature engineering, model retraining, and prediction.

Algorithmic Trading Leveraging Group Trends in Graph Representation

  • Organization: WUTIS - Academic Trading And Investment Society | LinkedIn | GitHub
    • As a leader of a 4-member team, achieved a first-place victory in the Algorithmic Trading pitch competition.
    • Created a graph representation based on the cross-correlation of stock price time-series data to identify group trends.
    • Backtested an algorithmic trading strategy based around stocks deviating and returning to group trends in the graph.

Pinned Loading

  1. the-impact-of-physicality-on-network-structure the-impact-of-physicality-on-network-structure Public

    Data and algorithms used in empirical contribution of the research project "The Impact of Physicality on Network Structure".

    Jupyter Notebook 4

  2. citadel_correlation_one_global_phd_datathon_2023 citadel_correlation_one_global_phd_datathon_2023 Public

    Welcome to my project repository for the Citadel & Citadel Securities and Correlation One PhD Global Datathon 2023 Fall

    Jupyter Notebook 1

  3. algo-trading-group-trends-graphs algo-trading-group-trends-graphs Public

    We applied #networkscience methods to #algorithmictrading in order to identify asset groups, which were implemented as a part of a #quantitativetrading strategy.

    Jupyter Notebook

  4. temporal-patterns-of-reciprocity-in-communication-networks temporal-patterns-of-reciprocity-in-communication-networks Public

    The study delves into the dynamics of human communication within various social settings, from intimate groups to global online platforms, focusing on the reciprocal exchange of information as a co…

    Jupyter Notebook 1