For our following analysis, we choose three cities San Francisco, Albuquerque and Detroit in following analysis demo.
In our exploration, we want to find the answers for questions that raised by business owners when they want to start or expand their business.
To start up, the business owner need to choose the city that they want to open their business. For this purpose, we are directed to the following question: which city should I consider?
In order to make better decision, an overview of city can help them get some insight, like the culture trend, the consumption level, the gathering places of restaurants. Here we use interactive bar plot to dig that hidden information.
Summary
By clicking the two buttons, we can compare different cities in various metrics.
- Culture Trend: choose tag. For example, compared with other two cities, Albuquerque has relatively higher bar of Mexican food. It indicated that people in Albuquerque like Mexican food more, which is true since it is close to Mexico.
- Consumption Level: choose price. By checking the price distribution, restaurants in San Francisco generally have higher price than other two cities. Business owners can set a little higher price in San Francisco.
- Restaurant Quality/Reputation : choose rating. For example, the distribution of the rating in San Francisco is light in the left tail and heavy in the right tail, which means restaurants in San Francisco rarely have low reputation and business owners can expect their future ratings high.
- Gathering Places of Restaurants: choose area.Take San Francisco as an example. From the plot, most restaurants congregate in SoMa, Misson and Financial District. Here we plotted Choropleth maps (Click to view). They are bustling areas and thus they are good choices for opening a new restaurant.
The city is a large area with various neighborhood districts. So business owners is curious about the business condition of each district, which can help them pin down the location. The interavtive boxplot is utilized to explore the information.
Boxplot for Rating/Price in Different Areas
The orange bar in the boxplot shows the range of mean to 75% quantile while the green bar displays the range of 25% quantile to the mean.
- Click the button of rating: we can see in San Francisco, restaurants in Bayview-Hunters Point, Bernal Heights, Castro, Misson, Potrero Hill, SoMa areas have higher rating than others, which means restaurants in these areas have higher reputation than others. Customers who go to these areas prefer to find a good restaurant. Starting or expanding a new business there, Business owners will face higher market competition.
- Click the button of price: take San Francisco as an example, restaurants in Chinatown, Inner Sunset, Mission, Outer Sunset, Tenderlion areas are cheaper than others. Therefore, business owners need to care more about other factors such as tasty and service if they want to set the price high in these areas.
After the city has been decided, the business owner may want to go deeper and narrow down the restaurants to certain category. The following questions are interesting to them.
What's the reasonable price for your business in the city you have chosen? And what rating do you expect to have for your business in the future?
We can know the distribution of price for each category in a city from the interactive bar plot of price. Also the histogram of rating tells us the distribution of rating for each kind of food.
Barplot for price
Histogram for rating
The relationship between price and rating is observed when we check the distribution of rating under different prices. While the price goes up, the rating also goes up. contingency table analysis is used to test this hypothesis.
H0: price and rating are independent VS H1: price and rating are dependent
| Rating Level | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Price Level | 1.0 | 1.5 | 2.0 | 2.5 | 3.0 | 3.5 | 4.0 | 4.5 | 5.0 |
| $ | 6 | 23 | 61 | 177 | 375 | 631 | 816 | 470 | 131 |
| $$ | 5 | 7 | 39 | 111 | 325 | 821 | 1027 | 408 | 132 |
| $$$ | 2 | 3 | 4 | 14 | 34 | 107 | 135 | 43 | 19 |
| $$$$ | 3 | 3 | 2 | 3 | 3 | 9 | 30 | 27 | 8 |
P-value is < 2.2e-16, so p-value is smaller than alpha and we reject the null hypothesis. We are 95% confident that price and rating are related with each other. Therefore, business owners should not lower the price to attract customers without considering other factors like rating.
Other Questions:
Which attributes play a major role when it comes to attracting more customers, like Wifi, Parking, ambience etc.?
-
Is wifi related to the price?
H0: wifi and price are independent VS H1: wifi and price are dependent
Contingency Tables for Price and WiFi Price Level WiFi $ $$ $$$ $$$$ No 1074 1253 163 39 Free 576 600 59 14 Paid 6 5 6 1
p-value = 4.585e-06 < 0.05, so p-value is smaller than alpha and we reject the null hypothesis. We are 95% confident that wifi is related to price. Therefore, if business owners plan to include wifi in their restaurant, they can raise the price properly.
-
Is wifi related to the rating?
H0: wifi and rating are independent VS H1: wifi and rating are dependent
Contingency Tables for Rating and WiFi Rating Level WiFi 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 No 0 14 38 113 344 785 917 294 45 Free 1 2 11 58 168 387 424 171 28 Paid 0 0 0 0 1 8 6 3 0
p-value = 0.4232 > 0.05, so p-value is larger than alpha and we cannot reject the null hypothesis. Hence, wifi isn’t associated with rating. Having wifi will not strengthen the goodwill of a restaurant.