Discussion 1: Arcelia
Pierson (2016) writes that cluster analysis using k-nearest neighbor (KNN) is often used in the retail business sector to produce value and understand data. One example of this is theft prevention technology that is advanced enough to analyze and act upon collected video to detect missed packages in grocery carts (Pierson, 2016). For example, these systems, without the need for human interference, are intelligent enough to identify items in your cart and add them to the final checkout without you having to scan them. While Amazon has yet to unveil the details of the technology being used at its Amazon Go stores (now being piloted worldwide), Tillman (2021) notes that the store uses “a system of cameras, sensors, and/or RFID readers to identify shoppers and their items” (Just walk out section). This system keeps track of items in carts and those on the shelves and does not confuse items when removed from the shelves and then placed back. While it is not entirely known if Amazon is using k-nearest neighbor analysis at these stores, based on Pierson’s examples, they may be using clustering analysis at some level.
As noted by, Srivastava (2018), some advantages of the KNN algorithm include its ability to output powerful and competitive results while being known as a simple classification algorithm. Furthermore, Srivastava (2018) writes that the algorithm is used in regression analysis where it uses averages instead of ranking nearest neighbors. Lastly, the KNN algorithm is easy to implement in R with just a single line, where other methodologies may be more difficult to code.
References
Pierson, L. (2016, March 26). Solving Real-World problems with nearest neighbor algorithms. Dummies. Retrieved October 31, 2021, from https://www.dummies.com/programming/big-data/data-science/solving-real-world-problems-with-nearest-neighbor-algorithms/
Srivastava, T. (2018, March 26). Introduction to k-Nearest neighbors: A powerful machine learning algorithm (with implementation in python & R). Analytics Vidhya. https://www.analyticsvidhya.com/blog/2018/03/introduction-k-neighbours-algorithm-clustering/
Tillman, M. (2021, March 4). Amazon go and amazon fresh: How the “just walk out” tech works. Pocket-Lint. Retrieved October 31, 2021, from https://www.pocket-lint.com/gadgets/news/amazon/139650-what-is-amazon-go-where-is-it-and-how-does-it-work
Discussion 2: David
Good afternoon class,
K-nearest neighbor analysis is a supervised machine learning method which is used to solve both classification and regression problems. While cluster analysis is an unsupervised method, it can be utilized in conjunction with k-nearest neighbor analysis to address business problems. “The KNN algorithm assumes that similar things exist in close proximity” (Harrison, 2018). As a real-world example, k-nearest neighbor classification can be used to identify all employees that either have an active retirement account with the company or do not have an active retirement account with the company, based on age. This would allow a company to predict whether an employee would opt into the retirement program based on age and show similarities between data points through proximity. K-nearest neighbor is mainly advantageous compared to other modeling techniques that have more assumptions given that the K-nearest neighbor method is easy to use, versatile, and does not require additional assumptions.
Reference:
Harrison, O. (2018, September 10). Machine Learning Basics with the K-Nearest Neighbors Algorithm. Retrieved from https://towardsdatascience.com/machine-learning-ba…
Discussion 3: Tyler
The stock market is a great example of where a K-Nearest Neighbor algorithm (kNN) can be used. Stocks values are often determined by the value of other similar stock options. Using information such as this, we can see that if the stock for a particular company goes up, it is possible that similar companies will experience a lift in their stock value. However, it is possible that the stock value for similar stick options may go down due to this competition so there is still some industry research and expertise that goes into these predictions. The kNN algorithm can also be utilized to find similarities between old data and new data which could also be helpful in predicting the stock market (Singh, 2018).
Regression algorithms can also be used to predict stock market price adjustments. One of the disadvantages that one might experience using regression in the stock market would be predicting the behavior of stock in a new company. Regression algorithms are at their best with a lot of historic data to back them up and since many companies don’t have that, regression falls to a great disadvantage. One might argue that you can use the historic data from other similar companies to some effect but that is what kNN is already trying to do and will likely do it better than a user trying to find a similar company.
References
Singh, A. (2018, October 25). Stock price prediction using machine learning: Deep learning. Analytics Vidhya. Retrieved November 5, 2021, from https://www.analyticsvidhya.com/blog/2018/10/predicting-stock-price-machine-learningnd-deep-learning-techniques-python/.
