Plant Data [Viz] + Recommendation Engine
Skills/Technologies Used: Python (webscraping, data cleaning, machine learning), Github, Tableau, Jupyter Notebook
This first part of this project uses Python to webscrape the Missouri Botanical Garden plant database in order to create visualizations analyzing plant characteristics. Chromedriver and Selenium were used to scrape each search result page within the plant finder tool for each individual plant’s URL. Another script was used to access each URL to scrape all the characteristics listed for each plant.
Developing a plant palette comes with experience and seeing what works in the field. As a practicing landscape designer, I've noticed that landscape architects tend to stick to a particular palette, and are sometimes hesitant to try new plants. To try to address this issue, I created a plant recommendation engine that would suggest similar plants they wouldn’t normally consider.
I cleaned and reformatted the data in order to perform a cosine similarity, which required encoding and normalizing all categorical features in order to do a pairwise comparison of vectors. I then wrote a function to return the top 5 similar results, with some parameters to filter out plants that are too similar (same plant, different varieties), as noted by the same common name.