Making the most of market basket analysis
1. Making the most of market basket analysis
Welcome back to the final lesson of this course! In the last 3 lessons, you took in a raw dataset of movie consumption data, transformed it to a transactional dataset, figured out the appropriate level of support and confidence, created rules and visualized them. Now, you're going to put it all together and learn the true impact of Market Basket Analysis on any business.2. Market basket in practice
Before going into the details, let's take a step back and remind ourselves what Market Basket Analysis allows us to do. Market Basket Analysis allows us to understand what customers or users are choosing. Most importantly, their basket of items is of great interest and can be very informative to a retailer. In addition to individuals transactions, information about customers or users can be added to the analysis. This could help segment or cluster customers according to their preferences. For instance, if we would have more detailed information about users, a retailer could further segment the customer base into more actionable groups of customers.3. What influenced yogurt ?
Let's start off by seeing how Market Basket Analysis drives value in the offline world. By looking at the groceries dataset, imagine we want to understand what items trigger the purchase of yogurt? What information can be extracted from the dataset? We call again the apriori function on the Groceries dataset with minimum confidence of 80% and set yogurt as the item on the right hand side of the set of extracted rules. The rules with the highest lift are shown here below. It seems that purchasing vegetables, butter and cheese triggers the purchase of yogurt - does it make sense to you? You will notice that most products in these rules are usual products customers tend to buy when visiting the grocery store. A retailer could think of placing these items close to each other in the store such that it makes the shopping experience of the customer more convenient.4. What did yogurt influence?
Now, let's take the problem differently and ask ourselves what products did yogurt influence. The only difference with the previous slide is that we set yogurt as the left hand side instead of the right hand side. By calling the summary function on the set of extracted rules, it turns out that the set of rules is empty with the current configuration of parameters. There are no rules for which the confidence is at least of 80%. One tip is that we should not reduce the confidence level just in order to find rules. A confidence of 50% means that you are 50% sure that the inferred relationship holds true. Hence, the confidence you have in the inferred rule is not high enough to take actionable measures.5. Let's find out recommendations for movies!
The last two slides allow us to get some recommendations with respect to a specific item. One could think of building a small recommender system based on the Market Basket Analysis. Now it is your turn, let's find out which movie you should watch next!Create Your Free Account
or
By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.