-
I'm Mateus Xavier from Brazil. I hold a B.Sc. in Chemical Engineering and a specialization in Business Processes and Statistics.
-
I am passionate about data science, statistics, and marketing. I am constantly working on projects in these areas. Explore them in my data science portfolio below!
-
My hard skills include Python (pandas, numpy, matplotlib, seaborn, scikit-learn, statsmodels), SQL, Machine Learning (classification, regression, clustering, and time series), Data Visualization, Market Analysis, Customer Segmentation, and Statistics. My soft skills include self-directed learning, effective communication, teamwork, quick adaptability, problem-solving, and critical thinking.
- Marketing Insights for E-Commerce Company In today's competitive e-commerce world, marketers face a big challenge: understanding customer behavior and optimizing sales strategies. To tackle this, I used techniques like customer segmentation (using RFM and K-Means) to group customers and tailor marketing efforts, exploratory data analysis to find patterns and trends, and market basket analysis to identify products that are often bought together for cross-selling opportunities. We also looked at how much we were spending on marketing and how that was affecting our revenue and customer retention. This helped us to make changes to our marketing strategies and to keep our customers happy. These methods help us to create targeted marketing campaigns, keep our customers loyal and help our business to grow.
- Client Segmentation - Clustering
- Sentiment Analysis of Social Media Posts: Predicting Positive, Negative, and Neutral Comments In this project, I replicated the analysis method that I regularly use in my job to evaluate NPS (Net Promoter Score) comments. By applying this approach, I aimed to classify the comments as Positive, Negative, or Neutral to ensure timely actions on the feedback received. To achieve this, I employed several machine learning models, including Logistic Regression, Naive Bayes, SVM, Random Forest, and Gradient Boosting. Among these, Logistic Regression with class weights emerged as the best-performing model, demonstrating superior accuracy and reliability in classifying the comments accurately. This analysis confirms that leveraging sentiment classification can effectively streamline the process of handling NPS feedback, allowing for quicker and more targeted responses.