Andrew Romanof Portfolio

The Match Outcome Predictor in Dota 2 is a comprehensive project designed to predict the results of Dota 2 matches using a machine learning approach.

To learn more about this project, please visit the github respository!

NewsFlash is a user-friendly social media platform developed as a school project using Flutter and Dart. Our goal was to create a simple and intuitive application that combines the power of Flutter's cross-platform capabilities with Dart's efficient programming language. NewsFlash offers a seamless and visually appealing experience, focused on real-time news updates and connecting users.

To learn more about this project, please visit the github respository!

In this school project, the aim is to explore Netflix data using R Studio. Utilizing tools such as group_by, summarise, mean, ggplot, filter, mutate, replace_all, and remove_all, various data techniques are applied to gain insights into user preferences and content performance. Through data filtering and analysis, interesting trends and patterns are uncovered. Visualizations created with ggplot enhance the understanding of the findings, presenting them in an engaging and informative manner. Join this exploration into the world of Netflix data, showcasing skills in filtering, analysis, and visualization using R Studio for this school project.

To learn more about this project, please visit the github respository!

In this school project, I implemented and compared several advanced association rule mining algorithms, including the Apriori algorithm, PCY algorithm, Random Sampling, SON algorithm, and the multihash version of the PCY algorithm. The primary objective was to uncover meaningful associations and frequent itemsets within massive datasets, and to assess the performance and effectiveness of each algorithm in this context. I conducted thorough experiments, analyzing various factors such as runtime efficiency, memory utilization, and the accuracy of association rule discovery. To evaluate the algorithms, I utilized two diverse datasets and explored different sample percentages (20%, 40%, 60%, 80%, and 100%) as well as support thresholds (1%, 2%, and 5%). I also paid specific attention to the number of false positives for Random Sampling, shedding light on its performance in the association rule mining domain.

To learn more about this project, please visit the github respository!

In 2023, I actively participated in the Advent of Code, successfully solving daily programming puzzles throughout the month of December using Python. This engaging experience not only allowed me to deepen my proficiency in Python but also required the application of a vast array of data structures and algorithms to solve each problem which taught me something new everyday.

To see the problems and my solutions, please visit the github respository!