Tutorial 4

Tutorial 5

Discuss and evaluate suitable techniques/methods being used in literature while performing the big data analytics on the following:
a) Market Basket Analysis.
b) Customer Churn Prediction Analysis.
Please support your discussion based on a research paper.

Example Answer
The big data analytics on Market Basket Analysis could help to
· Provide combo offers based on products being sold together
· Organize and place the associated products in the market
· Predict inventory stock and time-bound sales such as festive seasons
Techniques/Methods to achieve the above are
a) Association Rule Mining (give mapping of this method for your topic analysis — reason for proposing this technique–justification, rationale…..)
b) Clustering
c) Feature Extraction
d) Summarization techniques
e) Time Series Analysis
f) Regression analysis
g) Classification algorithms

Customer Churn Analysis
· Determine the customers who are at risk of leaving and if possible on the analysis whether those customers are could be retained
Techniques/Methods
a) Logistic Regression (give mapping of this method for your topic analysis — reason for proposing this technique–justification, rationale…..)
b) Multiple Linear Regression
c) Neural Network
d) Decision Tree
e) Fuzzy algorithms
f) Evolutionary learning algorithms

My Try
Market basket analysis: the name itself explain that the seller wishes to have insight of the purchaser wish list using big data processing; with the objective of putting more item into their basket. In psychology, that would mean that it is the intervention between a human needs and wants. The outcome of market basket analysis motivated the technique of association rule mining; which is to up selling or cross selling.

To start with, a typical supermarket is an entity that has inventory of wide variety of product to be sold to customer. In the old era, a customer would come to the supermarket with a purchase list to be complete. In order to expand customer “needs” in to “wants”, most of the supermarket has adopt association rule mining. Previous studies (Dongwon Lee et al., 2013), the exploratory of single receipt having different product. For example, a customer that purchase salmon would normally purchase bread or customer purchase bread will also purchase milk. This would make the association of different product thus making more cross selling possible.

Later, this could further expand in to how the inventory should be place in the supermarket (Ibrahim Cil et al., 2012). The technique introduced ensemble of association rule mining with multidimensional scaling. Location placement of product does not limit to increase customer experience but increase the product in customer basket. Both techniques show dairy product correlate with breakfast product thus by placing them nearer to each other would increase sale.

Another well-known big data analytic would-be customer churn prediction analysis. The objective of this analysis is to reduce the existing customer or previous turning down the new offer from seller, retaining the customer back to the seller. A Process mining technique is used in (Onur Dogan et al.2020) studies show that different path taken while visiting a shopping mall does not end with cashier (transaction complete). The technique open up a possibility to understand a “travel log” that stored in a big data storage, which is complex and recursive iteration. Possible sales improvement can be performed if the path and time visiting different aisles can be modify; such as reducing size of aisle. 

References
Dongwon Lee, Sung-Hyuk Park, Songchun Moon. “Utility-based association rule mining: A marketing solution for cross-selling” Expert Systems with Applications Volume 40, Issue 7, 1 June 2013, Pages 2715-2725. doi:10.1016/j.eswa.2012.11.021

Ibrahim Cil. “Consumption universes based supermarket layout through association rule mining and multidimensional scaling” Expert Systems with Applications Volume 39, Issue 10, August 2012, Pages 8611-8625. doi:10.1016/j.eswa.2012.01.192

Onur Dogan (2020) “Discovering Customer Paths from Location Data with Process Mining”, European Journal of Engineering Science and Technology, 3(1), pp. 139–145. doi: 10.33422/ejest.v3i1.250.
ps. first time submitting, still learning what to do.