Hi Sreenath,

When you use predictive analytics, there is a need for businesses to understand each and every row-case that takes place inside the business. Most cases convert the process into a coalition game where the dependent variable is the outcome and the independent variables are agents in that game working as a coalition. Taking these specific scenarios as you asked-

The case of B2B sales

• When a client is lost or won, you want to see how each factor contributed to the client's whole journey, whether lost or won. This uses shapely values which converts the whole predictive analytics into a coalition game.

The case of customer churn in digital Marketing

• Similarly, in digital marketing, you use shapely values to find the contribution of each channel in user-journey and maybe use a Markov chain to recommend the next best channel by the finding probability of conversion when the user moves another channel.

Hope it helps your query. Please reach out, if you need more information on how to use shapely values using python.

Sreenath Shaju

hi

1) https://arxiv.org/pdf/1909.08128.pdf

2) https://par.nsf.gov/servlets/purl/10053370

Hi Sreenath,

The following articles address the application of coalition games or formation and machine learning

1. Analysis of Coalition Formation in Cooperative Games Using Crowdsourcing and Machine Learning

2. Bayesian reinforcement learning for coalition formation under uncertainty

3. Modeling coalition formation for repeated games using learning approaches

4. Investigating reinforcement learning in multiagent coalition formation

Hope it helps 🙂!