The objective of this hackathon to develop a predictive model that determines the likelihood for an insurance customer to churn - to seek an alternative insurer or simply drop out of the insurance market altogether. In light of the current pandemic, churn prediction can be used to offer targeted support and tailored services to certain customers vulnerable to churning. This means more people can continue to be covered when they most need it most and insurance companies can be more efficient at serving and retaining their customers.
I merged the four dataset to one, I noticed that the 'ID' has duplicates so i droped the duplicates then took one.
I converted the categorical variables to numerical variables using a 'for statement', i also converted the datetime to year and months.
I used different Algorithms but Decision Tree Algorithm performed so well while twilking the hyperparameters
Our Target variable is the 'Lapse' column
The Data set can be found https://zindi.africa/hackathons/the-zimnat-insurance-assurance-challenge/data