BTMA 431:

Calgary Parties

All project files can be found at the following GitHub repo:

GitHub

Motivations

I thought it would be fun to rank each of Calgary’s quadrants against each other, to see which quadrant parties the most, which is very important to us university students as well. What else do we do when we’re done exams?
Obviously that is a bit difficult to do so I did the next best thing, I took business license data and focused on stores/restaurants with alcohol and cannabis.
I want Calgarians to see the city in a different light, and to see how many locations around the city you can go to get your party started.

Data Collection Process

Scraping the API from Open Calgary: Open Calgary

Found the data on Open Calgary of Business License Data.
Used packages httr, lubridate, and jsonlite to call from an API within R to get the data.
Filtered for only currently LICENSED businesses. No expired businesses.
After that, I filtered the license types looking only for alcohol and cannabis related licenses using dplyr.
Then finally, I used stringr to search the address column, and then to add a new column called quadrant;
- NW, NE, SE, SW

Visualizations

Legend:

Purple – Alcohol sales
Green – Cannabis sales

Initial views show us that Southern Calgary has the most alcohol and cannabis stores. First impression is that they party the most.

I would recommend downloading the PowerBI files from the Dropbox link above. From there you can scroll around Calgary, and check out your favourite spots!

Same findings as before, it looks like the Southern quadrants of Calgary dominate when it comes to partying (business license of Alcohol and Cannabis).
Interesting is how low the NW is to the four other quadrants, wonder if that has to do with Higher Income neighborhoods, or more family-oriented living in the North West.

Hypothesis Testing (Proportions)

Null hypothesis: Northern quadrants proportion of stores = city average

Alternative hypothesis: Northern quadrants proportion of stores < city average

For the two Northern locations, I did a less than hypothesis test. This was to test to see if their proportions were statistically less than the city average. Both tests gave us small p-values, allowing us to reject the null hypothesis, and go with the alternative hypothesis that the Northern quadrants do have a smaller proportion than the city average. Or more accurately, there is a lower number of places to purchase alcohol and cannabis in the North compared to the city average

Null hypothesis: Southern quadrants proportion of stores = city average

Alternative hypothesis: Southern quadrants proportion of stores > city average

In the Southern quadrants I went with greater than proportion tests due to what we saw in the visuals. Both the tests reject the null hypothesis, meaning that the Southern quadrants are truly the party areas of Calgary, or more accurately, there is higher number of locations to purchase alcohol and cannabis in the Southern quadrant compared to the city average.

Results/Summary

In the end, it appears that the highest density or best place to go to party in Calgary will be the Southwest.
We found this throughout our visualizations and our proportion tests.
Moving forward, I wonder if I should divide the city into more quadrants as shown in the next picture.
I wonder, like I said previously about the NW, if higher income areas or family-oriented neighborhood affects the amount of locations where you can purchase alcohol and cannabis.
In the end, my conclusion seems to lead to more questions!

R Script - Code

#### Ben LeBlanc Mini Project Presentation
library(httr)library(dplyr)library(lubridate)library(jsonlite)library(stringr)library(plotly)library(plyr)

url <-'https://data.calgary.ca/resource/vdjc-pybd.json?$limit=50000&$offset=0'
business.license.data <- httr::GET(url = url)
business.license.data <- content(business.license.data, as = "text")
busn.license <- do.call("rbind.data.frame", lapply(business.license.data, fromJSON))
# i want to change the address column to just include SW, SE, NW, NE. So I can do an analysis based on the four subsections# of calgary. I will also use the neighbourhood column, if the 4 quadrants are not enough.
busn.license$quadrant <- str_sub(busn.license$address , start= -2)
# Licensed - All requirements have been met, the invoice has been paid and this new business is currently licensed.# Renewal Licensed - This is a renewal licence that has been paid.# we are going to focus on these two types of licenses
busn.license <- busn.license %>% filter((jobstatusdesc == 'RENEWAL LICENSED' | jobstatusdesc == 'LICENSED' ))
busn.license$licencetypes <- as.factor(busn.license$licencetypes)
levels(busn.license$licencetypes)
# [3] "ALCOHOL BEVERAGE MANUFACTURER" # [4] "ALCOHOL BEVERAGE SALES (DRINKING EST/RESTAURANT)" # [5] "ALCOHOL BEVERAGE SALES (DRINKING ESTABLISHMENT)" # [6] "ALCOHOL BEVERAGE SALES (RESTAURANT)" # [15] "CANNABIS FACILITY" # [16] "CANNABIS STORE"
busn.license <- busn.license %>% filter((licencetypes == 'ALCOHOL BEVERAGE MANUFACTURER' | licencetypes == 'ALCOHOL BEVERAGE SALES (DRINKING EST/RESTAURANT)' | licencetypes == 'ALCOHOL BEVERAGE SALES (DRINKING ESTABLISHMENT)' | licencetypes == 'ALCOHOL BEVERAGE SALES (RESTAURANT)' | licencetypes == 'CANNABIS FACILITY' | licencetypes == 'CANNABIS STORE'))

busn.license$licencetypes <- ifelse(grepl(pattern = "ALCOHOL", busn.license$licencetypes), "ALCOHOL", "CANNABIS")
busn.license$licencetypes <- as.factor(busn.license$licencetypes)busn.license$quadrant <- as.factor(busn.license$quadrant)busn.license$comdistnm <- as.factor(busn.license$comdistnm)
# first we want to see the levels of alcohol per quadrant
busn.license.alcohol <- busn.license %>% filter(licencetypes == "ALCOHOL")
table(busn.license.alcohol$quadrant)
as.numeric(table(busn.license.alcohol$quadrant))
p <- plot_ly( x = c("NE", "NW", "SE", "SW"), y = as.numeric(table(busn.license.alcohol$quadrant)), name = "Where to Buy Alcohol via Quadrant", type = "bar") %>% layout(yaxis = list(title = 'Count of Alcohol Shops'), barmode = 'group')p

# we can also look via community

# as we look, it seems that the south side of the city has more places to buy alcohol/to get intoxicated.# now let's look at cannabis
busn.license.canna <- busn.license %>% filter(licencetypes == "CANNABIS")
table(busn.license.canna$quadrant)
as.numeric(table(busn.license.canna$quadrant))
p1 <- plot_ly( x = c("NE", "NW", "SE", "SW"), y = as.numeric(table(busn.license.canna$quadrant)), name = "Where to Buy Alcohol via Quadrant", type = "bar") %>% layout(yaxis = list(title = 'Count of Cannabis Shops'), barmode = 'group')p1# comparing the two it looks quite similar. The south quadrants of the city have the higher count of stores, while the NE is catching up and the NW falls behind.

quadrants <- c("NE", "NW", "SE", "SW")alcohol <- as.numeric(table(busn.license.alcohol$quadrant))cannabis <- as.numeric(table(busn.license.canna$quadrant))data <- data.frame(quadrants, alcohol, cannabis)
p2 <- plot_ly(data, x = ~quadrants, y = ~alcohol, type = 'bar', name = 'alcohol') %>% add_trace(y = ~cannabis, name = 'cannabis') %>% layout(yaxis = list(title = 'Count'), barmode = 'stack')p2

# now that we have visualed our question as to which quadrant of the city contains the most accessible alcohol and cannabis, lets put it towards a hypothesis test# does each quadrant have a similar number of shops to the city average, or is there a statistical difference?# for this test we will be using both cannabis and alcohol, not seperating it

city.mean <- mean(as.numeric(table(busn.license$quadrant)))

# now we will conduct 4 seperate hypothesis tests, to see if any quadrant of the city is statistically different to the mean of the city# for the 2 Northern quadrants I will test to see if they are less than average# for the two southern quadrants I will do a hypothesis test to see if they are more than average# more specifically a proportion test
table(busn.license$quadrant)as.numeric(table(busn.license$quadrant))[1]
nrow(busn.license)

NEt.test <- binom.test(x = as.numeric(table(busn.license$quadrant))[1], n = nrow(busn.license) , p = 0.25, alternative = "less")
NEt.test# P-value = 2.166e-08
NWt.test <- binom.test(x = as.numeric(table(busn.license$quadrant))[2], n = nrow(busn.license), p = 0.25, alternative = "less")
NWt.test# p-value < 2.2e-16

# now for the two southern quadrants I will do greater than tests, as we saw in the visualizations they had higher counts than the northern quadrants
SEt.test <- binom.test(x = as.numeric(table(busn.license$quadrant))[3], n = nrow(busn.license) , p = 0.25, alternative = "greater")
SEt.test# p-value = 5.419e-05
SWt.test <- binom.test(x = as.numeric(table(busn.license$quadrant))[4], n = nrow(busn.license), p = 0.25, alternative = "greater")
SWt.test# p-value < 2.2e-16
# all the p-values come back, and we can reject the null-hypothesis on each one of them.# So in the Northern quadrants in the city, we can say that they have less stores to purchase alcohol and cannabis,# while the Southern quadrants of the city have more stores to purchase alcohol and cannabis.
# in conclusion, the part of the city that parties the most, based off of visualization and hypothesis testing is the SW.

Google Sites

Report abuse