Problem 1: Data-Based Decision Making
Supermarket Product Placement
Suppose that we are responsible for managing product placement within a local supermarket. Our shelving units have 6 shelves each and are numbered from 1 to 6—with 1 being the lowest shelf and proceeding upward until the highest shelf is assigned the number 6. While there are many placement options that we should consider, we decide to look for any correlations between the row a product is placed on and its sales. Since we have our data stored in a data warehouse, it is easily accessible and responds quickly to our data request. Consider each of the following:
· What judgments can you make ...view middle of the document...
Example 2: Because we can track our customer’s purchases over time, we are not limited to analyzing individual “market baskets”. Instead, we are able to perform a richer analysis of their cumulative market baskets over time. Our data mining program discovers the following item sets with significant support:
Item Set 1
Battat Take Apart Toy
Amazing Airplanes children's book by Tony Mitton
Child's Pilot Hat
Stomp Rocket Jr. Glow Kit
Item Set 2
Bigelow Green Tea
Cozy Chamomile Nighttime Tea
The Smarter Science of Slim book by Jonathan Bailor
Item Set 3
Being George Washington by Glen Beck
Killing Lincoln by Bill O’Reilly
Original Intent by David Barton
How can you use this information?
Answer: We can use this information to strategically locate items in the store. Group these items together and you may increase sales. We can also use this information to predict sales as the prices change.
Example 3: We use our data mining program to analyze local buying patterns. We discover that when men bought diapers on Thursdays and Saturdays, they also tended to buy beer. Further analysis shows that these shoppers typically did their weekly grocery shopping on Saturdays. On Thursdays, however, they only bought a few items. How can you use this information? …for product placement? … for promotions?
Answer: If people do more shopping in Saturdays, then promote and position products strategically on that day.
Problem 3: Market Basket Analysis: Concept Tree/Sequence Analysis
A local home improvement store has performed data mining and has identified the following concept tree which corresponds to a deck. The concept tree defines a “deck” in terms of the products that are sold at the store. What does this allow you to do when analyzing market basket data?
Answer: This allows you to visualize and determine every piece that will be required to complete a deck
After we were able to build this concept tree, we also perform sequence analysis on our past purchase data and have found strong support for the following sequences of purchases.
Deck -> Outdoor Furniture -> BBQ
Deck -> Flowers/Landscaping
How can you use this information?
Answer: You can use this data for upselling and cross-selling
Problem 4: Decision Tree
In the past, your company’s loan department paid their staff to analyze the credit worthiness of individuals—i.e., to decide whether or not an individual should be approved for a loan. Over the years, a large amount of data was generated related to this process. The data set contains information about the applicant, about their loan, and whether the individual met their responsibility to pay the loan back with regular monthly payments. This data has recently been used by data mining processes to look for patterns that determine an applicant’s credit worthiness. The following...