In this project, unsupervised learning techniques (machine learning) were applied to segment countries according to 15 variables of infrastructure, basic services, economy, demographics, forestry aspects, population and connectivity. Using correlation matrices, principal component analysis (PCA) and hierarchical clustering. After the PCA, the main dimensions were interpreted and, with clustering, the resulting groups of countries with similar characteristics were analyzed.

A multiple linear regression model was developed to analyze GDP per capita per employed person (2017 prices), using five World Bank predictors. Countries were classified as developed and emerging through a graphical analysis compared to the global average. The goodness of fit of the model was evaluated, considering the coefficient of determination (R^2), p values, coefficients and the assumptions of linearity, normality, homoscedasticity and independence of residuals.

In this project, supervised learning (machine learning) was used to classify countries as developed or developing using WHO data. The k-nearest neighbors (knn) algorithm and seven predictor variables were used: life expectancy, health spending, infant mortality, schooling, GDP, homicides and HDI. The system was trained and evaluated to build a model that analyzes multiple dimensions to discern the relative position of countries in terms of development.

This project applied linear models to time series of air quality and meteorological variables from the Share station in 2018. After processing the data, selecting a 200-day sample and dividing it into 10 windows, linear models were adjusted taking ozone as the response variable. and predictors such as solar radiation, temperature, humidity, wind and precipitation. Assumptions of linearity, normality, homoscedasticity, independence of residuals and stationarity were verified. Finally, ozone predictions were made under different scenarios of environmental variables, providing insights into the factors that influence air quality.