Niko Nolte 2023-05-05
The Beer Connoisseur is a premium beer review site for beer enthusiasts. They have a magazine and they also have over 8 million readers annually. I find this topic very interesting because beer has been around forever. I have had the opportunity to try beer all over the world and it is something that I will always hold dear to my heart. There are so many different kinds of beer and so many different tastes. I want to understand what factors contribute to how well a certain beer is rated as well as the different styles and attributes that relate to a beer rating.
I will be answering this question by scraping data from the premium beer website that I have identified above. I will be comparing appearances, aromas etc. and how it affects the overall score. I am also curious to see what some of the variables mean like the variable “mouthfeel” and how that would effect the judges rating as well as the overall rating. It is important to note that the “beer name” column also has a little description, which gives the reader a little bit of background on each beer. All of the beers are different which makes this scraped data particularly interesting.
One of the things that we were required to keep in mind while scraping data off of websites was the ethical practice of doing the actual scraping. To be as ethical as possible, we have been instructed to 1). if there is a public API to use it, 2). I will create new value from the data that is scraped and not duplicate, 3). Where possible I will use a user agent that defines who i am and makes my intentions clear, 4). I will request data at a reasonable rate, 5). I will only scrape the data that I need and 6). I will site my sources and appropriately attribute the work of others.
I want to introduce my columns and what they mean so that there is no confusion to what I am talking about during my analysis.
With this first graph it shows how many times the Aroma score was counted at each integer of its appearance. The graph also shows how many counts of each of the scores there were. For example, if you look at the score of 22 it has the highest amount of occurrences in the scraped data around a total count of 90. Aroma is a category that is out of 24, and according to this data, no beer ever received the max score of 24 for the Aroma category.
For this next chart I was curious about what some of the Average Impressions were from three distinct reviewers that seemed to have a lot of given reviews. This graph is showing the average scoring of impression. The overall impression score is out of 20, and it looks like the reviewer who has the highest average impression is the user named Susan Ruud. This can be used to determine what each of these reviewers are likely to rate an overall impression in the future. If I wanted to pick a reviewer to review my beer, I would choose Susan since on average her overall impression score that she gives is higher.
I was also curious about the average rating that each reviewer attributed to see if there was a clear outlier in certain reviews. Fortunately, there was not any drastic outliers that I can see from the sampled 240 reviews that I gathered. As seen most of the average judges ratings were around 85-90 which shows that most of the beer was well liked by each reviewer.
For this next graph I wanted to see the distribution of how much variation that each reviewer has on the attributes: “aroma”, “flavor”, and “mouthfeel”. I found this box plot to be very interesting because you can see the variation in each of the reviewers. For example, it looks like flavor has the most disparity among the attributes which I find interesting. It makes sense since different reviewers have different preferences and acquired tastes. This could speak to the question that I wanted to answer from this analysis. That being that “aroma”, “flavor” and “mouthfeel” all have different effects with different distributions which ultimately contributes to the judges score.
For this last graph I wanted to see the relationship between the Judges rating and the impression score for each beer. As expected, the higher the initial impression, the higher the judges score. As seen there is a trend that can be attributed to the higher the higher the impression the higher the score as stated before. This can answer my question of what attributes contribute to a higher rating for a beer. First impressions seem to be very important.
In conclusion there are a variety of different variables that go into rating a beer. These attributes can determine popularity of a certain beer and I find this specific website to be very interesting. I hope everyone has enjoyed my analysis and feel free to check out the website! It is very cool!