Machine Learning Engineer | Data Scientist

github

Satwik Behera  • Peerlist

Replicate any folder of your Google Drive on your system (or wherever you decide to save the files).

Replicate Google Drive

A statistical investigation on the relationship between GDP and number of medals won in the Winter Olympics using the data sourced from Winter Olympics Medal Count dataset and Worldwide GDP History dataset. We performed a simple linear regression analysis with the weighted total medal points as our response variable and log(GDP) as our explanatory variable.

Does Money Buy Medals?

Cardiovascular diseases(CVDs) are the leading cause of death globally, accounting for 31% of all deaths, with an estimated 17.7 million lives lost each year. Early detection and management are necessary for people with heart diseases or high cardiovascular risk. Four out of five CVD deaths are due to heart attacks and strokes, and a third of these occur prematurely in people under 70. Since heart failure is common in cardiovascular diseases, we built a model that predicts heart failure which will help to predict a potential heart disease.

Heart Failure Prediction

We are living in a strange time, where we’re seeing an increase in the level of education throughout the nation, and yet the rate of labor force participation of the population age 25 and over is falling for all education groups. Many millennials, on a very large scale, are quitting their jobs, a phenomenon known as The Great Resignation, and yet others are doing a silent revolution, where they have restricted their participation to the bare minimum, a phenomenon known as Quiet Quitting.

I analyzed a dataset to look if it’s really the case that Millennials are more likely to quit, or is there some other confounding variable that is causing these changes in the work force.

This was my final project for the subject of STAT 650 - Advanced R.

Are Millennials more likely to quit their jobs?

Using Convolutions and Inception network (through tensorflow hub) to classify images into one of the six categories and comparing the training and validation accuracies achieved through each one of them

Intel Image Classification

This program takes an unsolved sudoku puzzle as an input and prints its solution applying the backtracking algorithm.



Sudoku_solver

A Machine Learning model for cash forecasting to be used in treasury management which can be done either from historical cash receipts or from historical account receivables data.

Cash Forecasting for B2B

Learning, Implementing and teaching OpenCV simultaneously, giving a beginner's perspective to both learning, and implementation

OpenCV_Tutorial

Optical Character Recognition on printed and handwritten text using Tesseract and OpenCV

Optical Character Recognition

An NLP Text Classification Problem where I worked on a dataset containing the Tweets of users who have applied the following hashtags: #coronavirus, #coronavirusoutbreak, #coronavirusPandemic, #covid19, #covid_19
Each tweet was labelled from either one of the following 5 categories, i.e., Very Negative, Negative, Neutral, Positive, or Very Positive
I used 4 approaches, ranging from the most basic word embeddings to the class leading BERT model and compared their accuracy on both training and validation set, reaching upto  98.16% and 84.01% accuracy respectively