leopoldoagorio / challenges

Mono-repo containing several challenges in a variety of Data Science and Machine Learning topics.

About me

Hi there! I am an Electrical Engineer and Physicist with a Master's Degree in Artificial Intelligence and Robotics. As a lifelong learner, I am deeply passionate about Robotics, Machine Learning, and Data Science. Connect with me via or email me at [email protected].

💪 Check out my repository where I've compiled exercises, Kaggle challenges, and other interesting things I've learned.

🎲 Fun fact: I enjoy playing the board game Go⚫⚪ in my free time. You can find me on online-go.com (OGS) challenging players from around the world!

What am I doing right now?

👨‍💼 As of June 2023, I am the Lead AI Engineer of SereLabs, a startup focused on using Computer Vision for Industrial Aplications. We are currently working in the wood-plying industry.

🎓 In 2022, I successfully obtained my Master's degree in AI and Robotics (EEMsC, UdelaR), you can read my thesis at colibri. We are trying to publish two papers with the results.

📚 I am a teacher and researcher in the Systems and Control Department at the Institute of Electrical Engineering (SyCIIE). My previous work includes projects on Aerial Robotics and Reinforcement Learning, and Data Science for Electrical Demand Estimation. I nave also collaborated with the MISEs group on an open-source machine learning project that focuses on structural analysis. Click here to check it out!

Some Languages and Skills

challenges's Issues

Generate statistics for neighbourhood: min price, max price, mean, median

First issues with Airbnb database

• Los nombres de los barrios en neighbourhood_group aparecen en diferentes formatos (Eg: Brooklyn BROOKLYN)
• Los precios aparecen como $, Dollars o USD
• Hay datos sin el Neighborhood marcado
• Si no tiene reviews, reviews_per_month es NaN en vez de 0

Compute metrics (precision, recall, f1) on the test data

Generate statistics for neighbourhood group: min price, max price, mean, median

Classification with 70/30 split

• Train three classification models
• Report precision, recall and f1-score metrics
• Define better model

Report of empty values in dataset

Discuss how to improve data completion for neighbourhood

Train regression, with 70/30 split.

Report train RMSE and train two different algorithms, deciding which performs better.

Compare regression and classification and draw conclusions

Compute test RMSE and reconsider model

Graphically represent prices

Study correlation between columns and the dependant "quality" variable

Load dataset, inspect and do EDA