top of page

OBJECTIVE - 4 

Predicting opioid-related Accidental Deaths in Connecticut State. 

It is used to predict the class (or category) of individuals based on one or multiple predictor variables (x). It is used to model a binary outcome, which is a variable, which can have only two possible values: 0 or 1, yes or no, diseased or non-diseased. 

In this problem, we are focused to predict the accidental deaths associated with drug overdoses in Connecticut State are opioid-related or not based on Age, Sex, and drugs prescribed to the patients. 

Approach : Logistic Regression 

 

Reason 

​In this problem, we only need to find out which accidental deaths in Connecticut are opioid-related or not. So, in this problem the dependent variable contains only 0 and 1. 0 means not opioid-related and 1 means opioid-related. 

Output 

  • We have initially fitted a binary Logistic regression model and the model output is shown below. 

Picture8.png

As a next step, we look into the VIF value corresponding to each variable in the model to check whether there is any multicollinearity issue or not.  When we observe the VIF values, all the values are below 2, from which it is clear that there is no multicollinearity issue, since VIF corresponding to each variable is very low. 
 

Picture10.png
Picture11.png

  • From this model, we got an accuracy of 83.3% for the train data and the model gives an AUC value of 83.4%.  This shows the model is Good enough to predict the Accidental Death is opioid-related or not.  The sensitivity and specificity are 0.6590 and 0.9413 respectively on test data and 0.6642 and 0.9289 respectively on train data. The AUC curve is shown below. 

Code 

  • Availability of data, materials, and code are upon request.

Conclusion 

  • When we fit logistic regression, the model gives good accuracy of 83%.  Which is enough to predict the accidental deaths.

bottom of page