Results
Entering this project, the question I wanted to answer was whether users of AirBNBs had identifiable patterns of speech that would identify a negative sentiment of units being located in gentrifying neighborhoods. Ultimately, though I found identifiable patterns of sentiment by neighborhood, I did not notice a pattern when it came to gentrifying neighborhoods. This could be for a number of reasons, including (but not limited to):
Users have a general pattern of positive and/or vague speech in reviews. When it comes to a complicated issue such as gentrification, they simply don't want to go there.
Visitors may be unaware of gentrifying neighborhoods that are on the later end of gentrification - this would be the case with neighborhoods like Germantown or Cleveland Park.
There may be references to the problem, but my model is not tuned enough to recognize it out of hundreds of thousands of reviews.
Coming Soon!
My next steps will be further analysis on what features make an AirBNB appealing. I would like to continue NLP analysis to identify user types (large groups, couples, business, etc), and identify which features are most appealing to each party. I will ultimately make a recommender function that AirBNB operators can use to identify:
What type of user would be most interested in their property?
What features and experiences are this user type most likely to expect and/or appreciate?
In addition, I would like to view this data through the lens of time. How have the rates of change shifted over time? Are some neighborhoods picking up more quickly, while others are slowing down? Could we use this to identify which neighborhoods are the next big hits for AirBNBs?