Varun Kumar R

I have 1.5 years of IT experience and have completed PGP in Data Science and Artificial Intelligence from IIT Madras. I am actively exploring roles in Machine Learning, Strategy and Analytics.

https://www.linkedin.com/in/varun-kumar-r-1aa25625b/

● Data Analytics                                                            

● Data Visualization
● Market Research                                                         

● Marketing and Sales Analytics.

● Machine Learning                

● Artificial Intelligence

• Python ( VS Studio Code)
• SQL
• Power BI
• MS Excel

GitHub Link - https://github.com/VarunKumarR1994

• NLP News Title Classification - Web Scrapping, Data Cleaning, Text Vectorization, Topic Clustering, Topic Labeling, Classification Model, Model Evaluation. Tools Used - VS Studio Code.

The news titles were scrapped from BBC website. Tokenization and Lemmetization were performed on the news titles. Text Vectorization was done using Word2Vec. Topic Clustering was done using K means clustering. The news titles were labeled manually based on the clustering. The model was evaluated using both SVM and Naive Bayes. The accuracy of the model was found to be 0.947 in both the cases.

• Singapore Real Estate ML Regression - Data Collection, Feature Engineering, Model Selection, Model Evaluation, Streamlit Web Application, Render Deployment, Testing and Validation. Tools Used - VS Studio Code, Streamlit, Render.

Singapore Housing and Development Board flat resale data was given to perform ML Regression. Exploratory Data Analysis was performed on the given features. Random Forest, Decision Tree and Gradient Boost algorithms were used to perform the regression. The dependent feature is Resale Price and the independent features are Lease Commence Date and Floor Area. The Coefficients of Determination were found to be 0.586, 0.553 and 0.582 respectively. 

Web application using Streamlit was created for the users to input the details of a flat and predict resale price. The Streamlit application was deployed on render to make it accessible for users on the internet.