Datamining / DatascienceTo perform data mining and knowledge discovery, you need data. Although we're surrounded by data, much of it can't be freely used because it's private, covered by licenses and terms of use that almost no one reads.This was precisely one of the obstacles I faced when looking for a thesis topic. I spent a long time exploring and discarding subjects because I didn't have data I could use freely. That's how I discovered Open Data, which, similar to free software, allowed me to: use, modify, and share it freely. My thesis is about artificial reading of handwritten numbers in open data from legislative elections in the City of Buenos Aires. Handwritten digit recognition is a topic that spans various areas of knowledge and research, including both machine learning and pattern classification. Its importance stems from the practical applications it enables, such as automating the understanding of the high volume of printed documents used by businesses, governments, and an ever-expanding society. Some key examples include automatic postal mail management, processing bank checks, manual data entry in forms, government records, and printed credit cards > all widely used applications. Resumen de la Tesis (link 1) | Resumen de la Tesis (link 2) | Link a la tesis (Spanish) I also explored social network analysis, and here I'm sharing a final practical assignment from a Social Network Analysis course at the University of Michigan during Fall 2013. I used a network of Argentine film actors. The project was based on IMDb; I analyzed films and actors, created a graph data structure, and analyzed the relationships between actors who worked together. This is a Social Network Analysis of Actors. |