Contributed by: Pramod Paluri
I am an Engineering graduate, passed out in 2002. I have been in the IT industry for 16 years. I am working largely as a Data Architect/Business Intelligence Architect across various domains. I have international exposure too – having worked in the UK, Australia.
I have been employed with a start-up, Microsoft being our main client. We are a Microsoft Gold partner.
Given the COVID crisis, and ever since I have been in the Data Science Program with Great Lakes Institute, I wanted to apply my skills in Machine Learning to try and help society, or in this case, my company. I could see the number of Covid +ve cases increase, people suffering, and people with existing conditions (like heart disease, diabetes) being more likely at risk of death. I wanted to help my organization make sure people with existing conditions or other vulnerable people are informed of the risk & take precautionary measures to avoid a serious situation.
Covid-19 is a new strain of the virus with a lot of unknowns. Scientists had little idea how it could impact humanity. It was important to understand what factors made people more vulnerable to becoming very sick due to Covid. From the business standpoint, it was important to look at business and humanitarian risk to the company. I thought it would be great to predict how severely sick a person in my organization could get if Covid infected him. I thought this could be a great preventive measure to avoid getting complications for people (if we could predict the Severity a person could get).
There was data available in the public domain (compiled by Govt Agencies) that had many attributes of a person and the level of Severity of health complications he could get if he/she were to contract Covid. The idea was to take a sample of this data, Train machine learning models – to predict Severity a person could get and then later use the most Optimal model to Predict Severity for employees in my company.
As we could see worldwide, an employee getting COVID had serious business implications for my organization too. Although people were working from home, there was still a chance of contracting COVID at public places. We did not have data to see what kind of Severity of health issues a person could get if he were to contract COVID. If a person were to get sick or lose his life – it would be a huge loss to the company and the family and society in general. It could also mean loss of business opportunities and loss of revenue for the company.
I used Jupyter notebook, Python to build machine learning models. I used an excel file to read the data.
I used data cleansing, outlier treatment, looked for correlations between variables, and then various Machine Learning libraries within Python, to build Prediction Models.
This was not a binary classification but rather a Multiclass Classification. I did not face an issue of Class Imbalance in this case. I used Naïve-Bayes, RandomForest, KNN, Logistic Regression for building classification models.
The data had existing information on personal attributes and what level of Severity he/she had – of covid complications.
The insights I derived was that – Age had a positive correlation with the target variable. As Age increased, severity level increased. Similarly, smoking/diabetes had a positive correlation with the target. People in the lower bracket of the age group were less likely to get severe complications due to covid. If they were non-smoking, Severity was even lower. BMI was also positively correlated.
The recommendation/solution that I proposed was that employees with existing health issues (like diabetes, being overweight or heart conditions, drinking habits, etc.) were at a very high risk of getting severe complications if they contracted covid.
So, they need to be advised to take extra precautions to avoid getting COVID. Stay home, apply a mask whenever they are out, and stop getting into close contact with a huge group of people.
The organization was very happy with the recommendations. On applying the recommendations, the vulnerable folks took the necessary extra precautions to avoid getting Covid. The most important point being people were informed of the risks they face. An informed individual takes precautions, which lead to taking care of employees’ health and increasing the company’s business productivity overall.
Applying Data Science@ Work helped me in contributing to well being of society and organization.
Also, the company was happy to see that I have additional skills in Machine learning, which could be an asset in future projects. So, in essence, I am an asset to the company.
If you wish to upskill, join Great Learning’s PGP Data Science and Business Analytics Course. For more such success stories, watch this space.
