Medical Insurance Cost Prediction and Analysis

The project involves capturing trends and information by visualizing the data and formulating mathematical syntax in calculating the approximate insurance cost.

Data preprocessing and
visualization

Exploratory data analysis using visualization and observation outliers handling.

Multiple Linear
Regression

Quantitative approach in formulating the mathematical syntax that allows individuals to compute insurance based on personal details.

Logistic
regression

Probability extraction of the classification model in order to determine the limits of insurance cost.

Data Analysis Summary Findings

Exploratory analysis of medical insurance data set which visualizes observations.

Insights extracted from data visualizations.

Majority of people who avail insurance are within the age bracket 20 to 50 years old.
Insurance cost have a positive correlation with increasing age.
Insurance policy costs more for individuals who do smoke.
A gradual increase in cost for from having 2 to 4 children and decrease for having 5.

Insurance cost prediction

Using linear regression to formulate the cost of insurance mathematical model considering the attributes provided by each individual.

Predictive Mathematical Model

With mean absolute error of 2660 and R-squared score of 61%, our mathematical syntax would be:

y = 220.60x1 - 204.26x2 + 72.78x3 + 369.42x4 + 15919.58x5 - -426.05x6 - -2555.33

Setting the cost expectation

With a correlation value 0.62, smoking causes a major drag in our insurance cost. With this in mind, we would like to know what are the minimum and maximum value of being a smoker in contrast to non-smoker.

Using logistic regression, we can say the following conclusions::

We can expect minimum cost of approximately 23,000 for being a smoker.
We can expect maximum cost of approximately 22,000 for being a non-smoker