I was actually expecting to see data that would refute the initial claim in the reverse! The outcome was not at all what I was expecting.
The assumption that I had made was, "More foot traffic = more customers = more reviews = higher overall rating". I was expecting to see a very high score close to train stations, and a hard slope the further you went from stations. My thinking was that location would heavily favor the rating, both in terms of convenience for customers, and in terms of non-Parians eating the food, because they are on vacation, traveling, etc. as the author had mentioned. I figured that if they are in higher spirits, they would leave higher reviews.
After seeing the results, it does make me wonder if this might have still played a role, but to a lesser degree? Hard to say, given how scattered it all is. Botting really does ruin metrics like this.
Personally, I think, it would be interesting to see something like the following in part 2:
1) Distance from station and proximity of kebab shop to bars/nightclubs. Perhaps someone who is drunk, and plans on taking the metro back to a hotel is likely to think a kebab is far better, compared to one who is further away from heavy drinking and further away from transportation. Both of those things being removed would make me think the overall review would be lower, but I think the actual ratings would be far more accurate, and more likely given by Native Parisians (assuming it's by a neighborhood, or whatever.)
2) It would be interesting to see what the impact is, in terms of amount of reviews reviews, the further away a kebab shop is from a station.
For example: If we are to assume that kebab_1 and keba_2 both open within six months of one another, but one is 1km from a station, while the other is almost on the platform, how much will that impact the number of reviews received?
3) Finally, it might also be interesting to hear what other food review websites that might offer this type of information. I assume the French have at least one French-centric social media platform for food reviews, which you can (hopefully) grab data from. How does that sites info compare to Google?
3a) Other nations food review sites might be interesting, too. My understanding is that Iran and Turkey are both very passionate about kebab, I could imagine them have thorough reviews of Parisian kebab shops. I could also see Japan having a pretty passionate food review site, given how crazy the Japanese can get for France, Paris in particular. I.e. https://en.wikipedia.org/wiki/Paris_syndrome
4) Perhaps there would be some interesting datapoints to pull out based on places that have closed permanently. That is, "Is a kebab shop that is close to a station more likely to stay open, or close down?"
The assumption that I had made was, "More foot traffic = more customers = more reviews = higher overall rating". I was expecting to see a very high score close to train stations, and a hard slope the further you went from stations. My thinking was that location would heavily favor the rating, both in terms of convenience for customers, and in terms of non-Parians eating the food, because they are on vacation, traveling, etc. as the author had mentioned. I figured that if they are in higher spirits, they would leave higher reviews.
After seeing the results, it does make me wonder if this might have still played a role, but to a lesser degree? Hard to say, given how scattered it all is. Botting really does ruin metrics like this.
Personally, I think, it would be interesting to see something like the following in part 2:
1) Distance from station and proximity of kebab shop to bars/nightclubs. Perhaps someone who is drunk, and plans on taking the metro back to a hotel is likely to think a kebab is far better, compared to one who is further away from heavy drinking and further away from transportation. Both of those things being removed would make me think the overall review would be lower, but I think the actual ratings would be far more accurate, and more likely given by Native Parisians (assuming it's by a neighborhood, or whatever.)
2) It would be interesting to see what the impact is, in terms of amount of reviews reviews, the further away a kebab shop is from a station.
For example: If we are to assume that kebab_1 and keba_2 both open within six months of one another, but one is 1km from a station, while the other is almost on the platform, how much will that impact the number of reviews received?
3) Finally, it might also be interesting to hear what other food review websites that might offer this type of information. I assume the French have at least one French-centric social media platform for food reviews, which you can (hopefully) grab data from. How does that sites info compare to Google?
3a) Other nations food review sites might be interesting, too. My understanding is that Iran and Turkey are both very passionate about kebab, I could imagine them have thorough reviews of Parisian kebab shops. I could also see Japan having a pretty passionate food review site, given how crazy the Japanese can get for France, Paris in particular. I.e. https://en.wikipedia.org/wiki/Paris_syndrome
4) Perhaps there would be some interesting datapoints to pull out based on places that have closed permanently. That is, "Is a kebab shop that is close to a station more likely to stay open, or close down?"