EVALUATION OF PREDICTIVE REGRESSION MODELS TO ESTIMATE SOFTWARE EFFORT
Effort estimation is a critical task in the software development lifecycle. Inaccurate estimates can cause customer dissatisfaction and reduce product quality. In this article, we evaluate the use of techniques based on machine learning to estimate effort for software tasks. We performed an empirical study based on the data set [Desharnais 1989] and compared the predictions of three models: Linear Regression (LR), Support Vector Machine (SVM) and K-Nearest Neighbor (KNN). The results of our study show that some software metrics are more important for estimating software effort than others. Also based on the quadratic error, which calculates how close the distance of a square regression line is to a set of points, we can successfully estimate 76% of the software effort for the studied data set, where the predictive models created show only 3% difference between them.
URBAN GROWTH PATTERNS RECOGNITION IN SATELLITE IMAGES WITH DEEP LEARNING AND EXTREME VALUE THEORY
The recognition of urban growth patterns in satellite images has applications that range from understanding the dynamics of urbanization to predicts future urban expansion. Both the availability of global inventories of land use, based on remote sensing, and advances in deep learning methods, offer an opportunity to boost the state of the art of existing models for this purpose. This task has broad implications for disaster preparedness, environment, infrastructure development and epidemic prevention, as well as developing new computer vision methods for time series data. Inspired by sequential models, this work proposes a method for detecting anomalies, or alterations, using the Peaks Over Threshold (POT) algorithm, a parametric probabilistic approach based on the Theory of Extreme Values that does not require manually defined limits and does not presuppose the data distribution. The algorithm was applied to representations obtained by a convolutional neural network (U-Net architecture) in order to recognize and detect possible changes in the geography of the regions, taking advantage of a temporal sequence of remote sensing images extracted from the SpaceNet dataset.