__STYLES__
About data set:
In 2017, Walt Hickey from FiveThirtyEight conducted a study to determine which Halloween candy is most popular among people. To do this, he created an experiment where online participants were asked to choose their preferred candy from randomly paired options (such as Twix vs. Snickers).
After collecting around 269,000 votes from 8,371 unique IP addresses, Walt calculated the winning percentage for each candy and analyzed various important factors (like whether the candy contains chocolate) to identify what makes certain candies more liked than others.
Objective:
The goal of this project is to analyze online voting data for 85 types of candy and identify the top 3 treats to distribute on Halloween. The aim is to ensure that trick-or-treaters of all preferences and dietary needs will find something they enjoy, with data-driven insights supporting the selection.
The report has been designed in a playful and engaging format, inspired by the Halloween theme.
By adding a slicer, I provided the ability to explore the various attributes of all the candies in more detail, allowing users to select their own top 3 favorites.
To satisfy a diverse group of trick-or-treaters, I identified the top candies in three key categories:
Health-Conscious Trick-or-Treaters: Selected the candy with the lower sugar content to cater to more health-conscious individuals.
Allergy-Friendly Option: Chose the top-rated nut-free candy to accommodate those with nut allergies.
Maximum Popularity: For trick-or-treaters without dietary restrictions, I selected the candy with the highest winning percentage in the rankings, ensuring broad appeal.
Additionally, I pay attension if there's a diversity of flavours to satisfy all.
Data Transformation:
Given that some data were initially unstructured and not optimized for analysis, using Power Query, I first cleaned and organized it to facilitate meaningful insights. Key transformations and groupings included:
Additionally, I created a table called “All Flavors” to list every flavor combination each candy contains.
For further analysis, I segmented the candies based on their sugar and price percentile:
Sugar Percentile Segments:
Low: less than 30%
Medium: 30%–70%
High: Over 70%
Price Percentile Segments:
Low: less than 30%
Medium: 30%–70%
High: Over 70%
After transforming the data, I created several DAX measures that helped me identify the top candies across three key trick-or-treater categories.
Data model view:
** Some data appeared to fall under unrealistic sugar percentile. These anomalies need to be investigated further.*
Key insights:
Top 3 candies that were choosen:
Flavours: Despite nut-flavored candies making up only 16% of the total selection (about 1 in 6), they are the most popular, winning 64% of the time. In contrast, fruit-flavored candies, which represent nearly half of the selection, are among the least favored options. Therefore, it's wiser to opt for nutty chocolates or nougat over fruity flavors.
Regarding candy texture, the most popular option tend to have a crispy texture, with caramel being the most favored flavor. Hard candies, as they are primarily fruit-flavored, are the least popular. The majority of the analyzed candies, however, are characterized by a chewy texture.
In general, the sweeter and more expensive the candy, the more popular it tends to be. Interestingly, many of the top-rated candies fall within the mid-price range. In the range of (60—70) price percentile, the average popularity (Win %) is over 20% higher than the overall average of around 50%. There are fewer options in the affordable (low-price) segment. The top three candies with the highest Win % in this category are: (1.)Reese's Miniatures (82%) , (2.)Starburst (67%) , (3.)Skittles Originals (63%).