finanCPLEX

Machine Learning and CPLEX Demo

This demo involves using machine learning and prescriptive analytics to promote financial products to banking customers.

Highlights

Data Exploration and Visualization

• Examine known behavior of customers
• Plot age, income, savings, pension, mortgage plots

Predict Customer Behavior

• Create a ensemble learning model using sklearn
• Visualize the predicted output

Compare Optimization Algorithms

• Write a "greedy algorithm" by hand
• Run a CPLEX solver in the cloud to compare revenues

Project-Specific Information

Tools Used

• Python, through Jupyter Notebooks
• Scikit Learn for model creation
• matplotlib for plotting
• DOCPLEX for CPLEX in the Cloud

Model Information

Gradient Boosting Classifer

• Response variable is probability of each product
• Features include age, income, members_in_household, loan_accounts

Relevant Files

Notebooks

• MachineLearning_CPLEX.ipynb

Data Assets

• known_behaviors2.csv
• unknown_behaviors.csv