Understanding customer perceptions of Chinatown restaurants (based on Yelp)

Introduction

Food in Chinatown, Manhattan has been increasingly popular since the turn of the 20th century, where like many other immigrant hubs e.g., San Francisco Chinatown, constructed their own semblance of “hometown” using local ingredients (Sietsema, n.d.). Today, Chinatown is renown for its elecletic food scene, a go-to for locals for all kinds of Chinese food (Grenier, n.d.). Personally, as an Asian American who grew up in the NYC area, I have also heard of Chinatown’s reputation as “cheap and tasty.”

However, to investigate whether its reputation is more than just “hearsay,” I will be using Yelp data. Specifically, do customer reviews on Yelp reflect Chinatown’s reputation in NYC as cheap and tasty ?

Some sub-questions I considered for this inquiry:

1. Are there significant differences in the Yelp reviews for Chinese and non-Chinese restaurants in the Chinatown area? Are there significant differences in their prices?
2. Is there a correlation between rating and price?
3. What are some reasons why a restaurant receives a low rating?

From Jupyter Notebook

The data was collected using the Yelp Fusion API. Yelp is a popular mobile app and website that publishes crowd-sourced reviews about businesses. To help filter down the data for this research question, I set specific conditions in the available Yelp parameters:

#to query all restaurants in the chinatown area
params = {
'limit': 50,
'location': 'chinatown, newyork',
'term': 'restaurants'
}

#to query chinese restaurants in the chinatown area
params_chinese = {
'limit': 50,
'location': 'chinatown, newyork',
'term': 'restaurants',
'categories':'chinese, All'
}

For the location, it was important to query for New York since there are other “chinatowns” in other cities. Additionally, since there are other stores in chinatown and we are investigating food specifically, I queried for restaurants only. Lastly, because we are identifying reviews made through the Yelp platform, I queried for chinese restaurants based on how Yelp categorizes Chinese restaurants instead of self-categorizing from the “all restaurants” dataset. (Note: category was determined by dictionaries provided on Yelp).

Findings

Understanding the Case Study Site: Chinatown, NYC

Before delving into the findings, it is helpful to understand what we are investigating. So far, we know that we are looking at restaurants in Chinatown, NY, but because the area is a melting pot, there are other categories of restaurants beyond Chinese food.

I looked into the different types of food available in Chinatown as well as which types are most prevalent:

There were 90 total categories ranging from Bakeries to Noodles to Tapas. However, the most prevalent restaurants types were Italian, followed by Chinese and American (New) respectively.

print(df_chinatown['category'].value_counts())

Italian                 51
Chinese                 48
American (New)          25
Dim Sum                 15
Seafood                 14
                        ..
Hong Kong Style Cafe     1
Poke                     1
Halal                    1
Georgian                 1
Salad                    1
Name: category, Length: 90, dtype: int64

Even among the Chinese restaurants, there were multiple types (27 total). Categories ranged from Szechuan to Tea Rooms. However, the most prevalent Chinese restaurants types were Chinese (general), followed by Dim Sum and Cantonese respectively.

Overall, Chinese restaurants represent about 43.2% of all restaurants in the Chinatown area, which is almost half!

Now that we understand the data analyzed,
Do customer reviews on Yelp reflect Chinatown’s reputation in NYC as cheap and tasty ?

Findings found that customer reviews on Yelp supported Chinatown’s reputation in NYC as being cheap and tasty. On average, restaurants received a rating of 4.11 (out of 5) with an average cost of 2.0 “$” or “$$” (out of 4 dollar signs). When filtered to Chinese restaurants specifically, restaurants received a rating of 3.68 with an average cost of 1.6 “$”. Thus, although the average rating for Chinese restaurants were slightly lower, there were also slightly cheaper.

Nevertheless, using the chi-squared test, there were no significant differences in the ratings for Chinese and non-Chinese restaurants
(p > 0.05). We can also see in the graphs below that were no significant differences in ratings and price between the restaurants based on its type.

Average Rating & Price (based on Yelp) for Restaurants in Chinatown

Note: categories were recategorized as chinese vs. non-chinese based on Yelp categories. Ambiguous / other represent categories that can capture both chinese and non-chinese restaurants e.g., speakeasies. This was a limitation in Yelp data such that there was no way to identify whether a restaurant belonged to a specific restaurant type without going through each individual restaurant.

Are these results driven by rating and price? Based on a Pearson r correlation test conducted between rating and price, the strength of correlation is low (r = 0.01), such that if rating increases by 1, price only increases by $0.01. Therefore, price does not fully account for the difference in rating.

So, what are some reasons why a Chinese restaurant receives a low rating?

Some of the factors listed in the reviews of the worst rated Chinese restaurants in Chinatown include: location, quality of food, service, and food hygiene. It is important to note that a limitation in Yelp queries limits collection of reviews to individual businesses, such that reviews can only be pulled for one business at a time. Therefore, the factors may not represent the full picture of reasons for a poorly reviewed location.

Conclusion

Although the findings supported Chinatown’s reputation as “cheap and tasty,” it did not necessarily mean that the Yelp ratings for its restaurants were high. Ratings included factors beyond just taste e.g., service, etc. Additionally, findings suggest that Chinatown’s reputation applies to all restaurants in the area, not just Chinese restaurants.

For future research, other crowd-sourcing apps can be beneficial to overcome the limitations of Yelp queries and or more extensive manual research on Yelp reviews needs to be conducted to understand the nuances in rating differences. Further research on Chinatown’s reputation may also help explain the reasoning beyond ratings i.e., the location has been historically known for crime and drugs and can have an negative impact on the ratings.