Harshita Harish

Analytical and adaptable data science graduate with hands-on experience in Python and multivariate regression. Delivered actionable insights on education trends at IdeasUnbound and communicated results through Tableau dashboards. Strong team player with a focus on data-driven problem solving.

Interactive Data Visualization – Coal’s Global Legacy
Tools: Python, D3.js, HTML/CSS, JavaScript

• Developed an interactive dashboard visualizing coal production, financial backing, and methane emissions worldwide
• Integrated and cleaned multi-source datasets (JSON, CSV) using Python for compatibility with D3.js visualizations
• Designed dynamic map, network, and tree-based visuals with drill-down filters and tooltips for policy insights

Beauty Industry Inclusivity Analysis
Tools: Python, R, Kaggle Datasets, SQL, CRISP-DM

• Conducted data-driven analysis on foundation shade diversity across global brands (e.g., Fenty, Il Makiage)
• Performed statistical modeling and sentiment analysis on customer reviews from Sephora and Nykaa
• Proposed a recommendation framework to optimize shade offerings using shade hex value clustering and regression

Happiness Prediction using Regression Models
Tools: R, Kaggle, Stepwise Regression, MLR, BIC

• Built and evaluated multiple linear regression models to predict happiness from 70+ survey predictors
• Applied bidirectional stepwise regression (BIC-based) to optimize model complexity and RMSE
• Achieved RMSE: 6.67, placed 60/231 in class Kaggle leaderboard
• Achieved private score: 0.3659, public score: 0.57825, ranked 80/231

Spam Email Detection using Classification & Clustering
Tools: Python, Scikit-learn, KMeans, Logistic Regression

• Engineered features from raw text and built ensemble classification models for spam prediction
• Integrated KMeans clustering to enhance feature separation for improved performance