Welcome to my data science projects portfolio! Here you will be able to check out some of data science projects I've made! The projects' main goal is to demonstrate my habilities to solve real business problems with Data Science.

I'm a data scientist who has been practicing data science skills since early 2020. My main focus is to create end-to-end data solutions for business problems through collecting, processing, analysing data and through implementing machine learning models that help improve business performance.

Tools and softwares:

• Data Collection (SQL, MySQL, Postgres): ⭐ ⭐
• Data Processing and Analysis (Python - Pandas, Numpy, Seaborn, Pyplot, Matplotlib, Scikit-learn): ⭐ ⭐ ⭐ ⭐ ⭐
• Data Visualization (Power BI): ⭐
• Machine Learning Modeling (Regression, Classification, Clustering): ⭐ ⭐ ⭐
• Machine Learning Deployment (Heroku, AWS): ⭐ ⭐
• Development (Git): ⭐ ⭐

In this project, I created a XGBoost model that predicted the TopBank customers' probability to churn and also formulate a action plan to tackle the churning problem based on giving customers a gift card in accordance to their churn probability and the maximization of customers' ROI. In addition to the financial return, the model was created using Dash and deployed in production with Heroku.

The model had a 0.905 F1-Score and can display how much revenue TopBank could save avoiding churn with gift cards, depending on the budget designated by the user. To check out the app, click here.

In this project, I created a XGBoost model that predicts the sales for the next six weeks. The sales forescating served as parameter to the budget designation for the stores' infraestructure renovation.

The model had a 763.11 MAE and the sales prediction per store can be easily access by Rossmann CFO through a Telegram bot. To access the Telegram bot, click here.

In this project, I created XGBoost models capable to predict 5 assets (Dollar, Bova11, Smal11, Bitcoin, Ether) rentability, and used Monte Carlo Simulation to select the best optimized portfolio.

In a case where the client intends to invest 1,000,000, using the rentability prediction of each asset's model, together with the optimization of the Monte Carlo Simulation portfolio, the client can expect to have a balance amount of 1,359,114.8 (35,89% return), so that the worst case scenario is where the balance is 1,116,539.8 while the best case scenario is a balance of BRL 1,601,689.8 (+- 17,78% volatility).

In this project, the main goal is to create a data solution through clustering, designed to create a customer segmentation that will orientate the banking marketing strategy.