2003 Federal University of Uberlândia (UFU)
Graduate Degree in Economics
I’m Pietro and Alice’s father, as well as Maressa’s husband. I come from a poor home and had to start working at the age of 14. However, I always believed that education might improve my financial situation, therefore I never gave up. I spent my whole childhood in public school, and at the age of 20, I took a difficult decision: to enroll in two concurrent higher education courses (full-time economics and accounting sciences at night). Because my parents couldn’t financially support me while I finished my studies, I did all I could to earn some “exchanged”: scientific initiation scholarship, disciplinary monitoring, consultant of the junior firm of the accounting sciences course, and, most importantly, computer technician. The procedure was difficult, but I was able to complete both.
Perhaps because I was interested, restless, and intrigued by challenges, I opted to pursue an academic career in the third period of my two graduations. I passed various master’s degree selection procedures and selected to do the master’s degree in Administration at UFU since I still had to finish the accounting sciences course at the same university and was going to marry Maressa. I received a scholarship for my master’s degree and had a notable performance: to this day, stories are repeated about how one student managed to produce 35 papers in two years! Secret = a passion for statistical analysis and a willingness to assist everyone who approaches him.
At the age of 24, I was already an university professor, and while working on my master’s thesis, I attempted to expand my understanding of quantitative methods with a focus on applied statistics. At the same time, I completed the PhD process in administration at FEA/USP and the public contest for professor at the Federal University of Viçosa (UFV). I went every week for two years, 12 hours to go and 12 hours back, to complete the PhD disciplines in São Paulo while also teaching at UFV. In 2008, I won a public contest for a permanent teaching position in my hometown and university, and in 2011, I completed my PhD in Finance at FEA /USP with honors: around 40 published articles and thesis recognized worldwide, as well as the KPMG /IBEF Revelation Award in Finance and several media reports in Brazil. After all, it was a very contentious topic: I demonstrated that credit rating may be improved using psychological measures!
I was “trained” in the culture of data modeling. I don’t think it could have been any other way, because the concept was to be a financial researcher from a young age, with the goal of extracting knowledge about the nature of the relationship between response variables and explanatory variables.
So, over the last two decades, I’ve evaluated hundreds (yes, hundreds of articles and consulting reports!) of databases using classical statistical approaches such as Descriptive Analysis, Bivariate Analysis, ANOVA, Regression Analysis, Multivariate Analysis, Generalized Linear Model, Generalized Estimating Equation, Generalized Mixed Model, Time Series Analysis, Structural Equation Modeling, Spatial Analysis, Item Response Theory, and Computational Simulation.
The Econometrics discipline was my foundation for the first 15 years; however, in the last five years, I have focused on the Psychometrics discipline, due to a high demand to work with latent variables and validate measurement instruments, and, of course, because I have fallen in love with psychometrics techniques. Anyway, I’ve always worked with health experts during this time, so I’m comfortable with their procedures and language. In this context, I attempted to master a variety of proprietary software, including Excel, SPSS, Stata, Eviews, Amos, SmartPLS, and Statistica, as well as open-source software, including R, JASP, jamovi, GPower, and GeoDa.
However, in recent years, especially after 2020, I have focused my efforts on algorithmic modeling culture in order to work with massive databases and focus on prediction. As a result, I have worked to improve my understanding of the following machine learning techniques: Lasso and Ridge Regressions, KNN, Random Forests, Bagging, Boosting, Neural Networks, and Support Vector Machines (SVM). Other subjects in this culture (Linear Regression, Logistics and Stepwise, Decision Trees, Discriminant Analysis, Cluster Analysis, and Principal Components Analysis) are familiar to me because they are already widely used in the data modeling culture, and I have worked on them in dozens of situations.
In this context, I have attempted to study the most popular Data Science tools, including SQL, R, Python (numpy, pandas, matplotlib, seaborn, statsmodels, scipy, scikit-learn, and so on), PowerBi, and Tableau. I’ve used RStudio and Jupyter as IDEs within the Anaconda environment. But isn’t that the least of this information? As I frequently emphasize, tools are ephemeral!
I frequently tell this story… I spent five years using proprietary structural equation modeling software that no longer works for me. I performed a lot of research and decided on a free one that would initially fit me. It took me an afternoon to master this new software: I knew what all the small buttons (choices) I had to hit (take). First and foremost, comprehend the techniques!