rajajivignan/eda-potential-insurance-customers

Exploration of the potential customers of insurance. Also, drawing insights with the help of univariate and multi-variate analysis

EDA-Potential-Insurance-Customers

About Data

The Dataset is about the potential Insurance Buyers. The purpose of this project is to predict whether a person is willing to buy or not. The dataset is made possible with the resources provided in a Hack-a-thon (Analytics Vidya). Thanks to them for the opportunity.

More info about the data is found in data.md

Initial plan for Data Exploration:

1. Removing the unnecessary features first.
2. Replacing the null values and Data cleaning.
3. Data wrangling: mean, std, max and, also correlation
4. Univariate analysis: to understand single feature precisely.
5. Bi-variate and Multi-variate analysis: for drawing out insights, inferences.

Data Exploration

Univariate Analysis:

Understanding the Recovered Insurance types. By the count plot, it's clear that there are more individual account in the dataset.

After that, it is necessary to analyze the distribution of the customers among the cities mentioned with the help of a seaborn countplot.

In the process, the distribution of the policy premium recovered so far is found out to be right skewed.

Multi-variate Analysis

Explored the the policy type and their percentages at which the customers bought while comparing them whether the customer is individual or joint account holder

Also, found out the goldilock zone of the duration of policy holders to buy another policy.

A compilation of different plots using seaborn pair plots

The detailed insights from EDA is mentioned in the notebook.