Results driven-data scientist with expertise in Data wrangling, Statistical analysis and Machine learning. With a strong background in Industrial Mathematics and Economics, I am an agile project leader with a passion for AI and a track record of delivering impactful solutions and achieving business objectives.
Analysed customer's behavior and applied clustering models in ML to refine analysis and segmentation core of marketing campaigns across retail stores in Europe and South America countries.
This project aimed to understand customer segments and reviews to provide actionable insights for improving the e-commerce platform's services and customer experience in Brazil. After a comprehensive analysis involving data cleaning and preprocessing, I utilized K-means clustering to segment customers based on their behaviors and interactions. Additionally, I employed Natural Language Processing (NLP) techniques to categorize and analyze sentiment from customer reviews.
A personal project whereby I researched and analyzed chunks of sentiment reviews of Twitter users. Machine Learning classification models were used to determine a user expression such as happy, angry, sad, neutral and so on through their tweets.
Built a deep learning model to detect and classify facial expressions such as happy, angry, disgust, etc using the Convolutional Neural Networks algorithm for integration in company's website
A personal project designed to statistically analyse based on some general chemical properties in winemaking, the effect of these properties on the quality of the wine.
Utilized python to curate 98% data from Suicide watch and depression posts, built and optimized NLP algorithms to demonstrate a remarkable 72.1% classification accuracy with Logistc regressions in order to give valuable insights to assist in mental health research and support initiaties.
Developed a predictive model to identify bank customers likely to churn using machine learning and deep learning techniques. Preprocessed data, conducted exploratory analysis, and addressed class imbalance using SMOTE. Achieved balanced F1-scores and improved accuracy, providing valuable insights for customer retention strategies.