Inspiration
As we looked at the outfits we began to think about what it was that made these outfits unique. In the end, we came to the conclusion that we could calculate the affinity of the outfits that shared a group.
What it does
The application uses datasets based on the affinity between Color, Material and Garment Type characteristics to generate unique outfits that do not exist in the original data.
How we built it
We have used a Python programming environment together with data processing libraries such as pandas, sklearn and matplotblib, as well as Streamlit for the web user interface.
Challenges we ran into
Not having specialized in DataScience, it was a challenge to think of a way to process the data and draw conclusions through pattern recognition. Since we do not have advanced knowledge of clustering, we opted for heat maps through correspondence matrices.
Accomplishments that we're proud of
The first challenge in addition to having reached this point has been to generate a good algorithm that has met the expectations of the challenge.
What we learned
We have learned to deal with large volumes of data and to work with libraries such as pandas and matplotlib.
What's next for Affinity Mango Wardrobe
Had we had more time we would have further refined our association metrics, in addition to providing a richer interface.