Analytics Projects

Wine Retailer Email Marketing Evaluation

Goal: Evaluate whether an email promotion is effective & who to target with email campaign
Data: Experiment data on email promotion, customers’ past purchase behavior data
Methodology:
1. Use linear regression to evaluate the average causal effects (ACE) of email treatment on purchase;
2. Perform slice and dice analysis to evaluate the evaluate the conditional causal effects (CCE) of email treatment on purchase;
3. Build a causal forest model to predict the individual-level customer purchase-lift caused by email campaign

Key Words: R, Regression, Causal Forests, Slice and Dice, Incrementality, Experiments, Email Marketing

Goal: Develop a predictive model and a decision support system (DSS) that evaluates the risk of Home Equity Line of Credit (HELOC) applications
Data: An anonymized dataset of HELOC applications made by real homeowners, provided by FICO
Methodology:
1. Use Maching Learning approch to build a predictive model to assess credit risk, models tested including Logistic Regression, Decision Tree, Random Forest, KNN, Linear Discriminant Analysis(LDA), BaggingClassifier, AdaBoostClassifier.
2. Design and develop a running prototype of an interactive interface that sales representatives in a bank/credit card company can use to decide on accepting or rejecting applications.

Key Words: Python, Machine Learning, Logistic Regression, Decision Trees, Random Forest, KNN

Goal: Provide product portfolio revitalization recommendations to a toy horse company
Data: Conjoint data of 12 product profiles responded by 200 individuals
Methodology:
1. Use Regression to estimate the conjoint model at the individual level ;
2. Conduct Benefit Segmentation via Cluster Analysis of Conjoint Part-Utilities;
3. Conduct a priori segmentation using the variables gender and age;
4. Simulate market shares for different product-line scenarios.

Key Words: R, Conjoint Analysis, Cluster Analysis, Benefit Segmentation, Market Simulation

Goal: Design and create a data warehouse for Blue Hill Country Club’s database
Data: Membership and transactional data related to one year of operations of Blue Hill Country Club
Tool: MySQL
Steps:
1. Build two versions of data warehouses both containing relevant membership and revenue data, one in long shape and the other in wide shape
2. Create ad-hoc analytics queries using the data warehouse built in the first step to answer relevant business questions the management team may be interested in.

Key Words: SQL, Data Warehouse, ETL, Database Management, Ad-hoc Querying