Instructions for setting up a GPU instance on Google Cloud

For my third project I tried to predict travel destinations of new AirBnB users based on demographics, like age and gender, sign up behavior, device type and web session records tracking each interaction on the platform. After playing around with different features and trying out various classification models I ran into problems like class imbalance and data leakage. Subsequently, I decided to settle on predicting if a new user is going to book or not.

After our first project we dove into machine learning for the first time and learned all the different models for linear regression. This week I finished my second project in which I put all the new knowledge about linear regression to use.
My objective for this project was to predict the current market value of players in the top five European soccer leagues based on performance and career data. I was able to get insight into which factors have the most impact on the market value to if players are over- or underrated. After scraping data of around 5000 players from the web I came up with 14 features for building my model:

• Age, height, nationality, position, division 1 or 2
• Games, goals, assists, played minutes this season, games for national team
• World Cup or Champions league winner, in team since, expiry of contract, strong foot, total transfer proceeds

After arriving from Germany at the beginning of March, the bootcamp finally started this Monday. It was an very intense and long week packed with a lot of interesting stuff to learn. As expected, the lectures are high paced which means we have to practice a lot independently and apply it in our projects/weekly challenges. Practicality. This is the main reason why I chose Metis - you learn the right tools in order to apply them in the real-world and solve interesting data science problems. In our first week we immediately dove into pandas, git, matplotlib/seaborn, debugging, some more pandas and complexity theories.