Mruga Parikh

** Participation in Kaggle Competitions ** 1.) House Prices: Advance Regression Techniques --- Top 19% 2.) Predict Future Sales --- Top 47% 3.) Titanic : Machine Learning from Disaster --- Top 23% 4.) Predicting Google Play Store App Ratings Working as Data Scientist in Philips HealthCare. Ability to mine hidden insights located within large data sets of structured, semi structured and unstructured data.Understanding the project objectives and requirements from a business perspective and then converting it into data mining problems.Able to influence the growth of the company by identifying the opportunities in large, rich data sets and creating and implementing predictive models that leads to growth in revenue and profit. Working with SQL to prepare big data by performing joins and filtering on the data sets.Using CART (Classification and regression) algorithm for solving data mining problems in R. Much functionality provided by this package handling preprocessing techniques, near-zero variance predictors, achieving parallelism using CART. When handling with classification problems, decision trees and random forest is used to for predictive classification modelling, helping us interpret the output as probabilities and labeled classes. Evaluating the model based on confusion matrix, ROC curves cross validation techniques. Handled multi-label classification using neural network and k-nearest neighbor’s algorithms.