Medizen

Medizen mobile application was developed during a Lambda School Build Week and was my first experience as a Data Scientist working in a team. Our team had members from Web, Data Science and User Experience. Different members of the team were focusing in different parts of the application such as UI, Front-End, Back-End, Data Engineering, Machine Learning Engineering, UX. In this team, my main focus was Data Engineering, building and testing the Data Science API. To balance the tasks and help the team, I ended up working on the Machine Learning side of the project as well as the API. Our objective was to create a model for the Medizen app that could provide recommendations of cannabis strains based on users preferences.

We found a dataset of 2K+ treatment options composed by strain, type, rating, effects, flavor and description. Speaking with the team, we thought we could give the user the option of choosing the desired effects based on filters.

We engineered a new feature that was the combination of type, effects and flavor for each strain. Using Natural Language Processing with a K-Nearest Neighbors model, the users preferences was turned into a string that could be compared to this new feature, finding the most similar strains. The model was finished, pickled and used in the API which was tested with Unittests and Advanced Rest Client. Unfortunatelly, the mobile application Front-End was not finished, still we had access to the prototypes being able to speak about the user flow and functionality of the application.

The Zen part of the application is related to the ability the user has of creating a tretament plan, taking the time to medicate with some background music and then evaluate the effects to check if it is a good option treatment.

I learnt a lot about working in a team as a Data Scientist. I believe we had a good communication and good workflow and our DS API was able to deliver the expected results.

The DS API GitHub repository can be found here.