A solution for estimates in software development projects

The Corporate world is becoming more and more competitive. This leads organisations to adapt to this reality, by adopting more efficient processes, which result in a decrease in cost as well as an increase of product quality. One of these processes consists in making proposals to clients, which necessarily include a cost estimation of the project. This estimation is the main focus of this project. In particular, one of the goals is to evaluate which estimation models fit the Altran Portugal software factory the most, the organization where the fieldwork of this thesis will be carried out. There is no broad agreement about which is the type of estimation model more suitable to be used in software projects. Concerning contexts where there is plenty of objective information available to be used as input to an estimation model, model-based methods usually yield better results than the expert judgment. However, what happens more frequently is not having this volume and quality of information, which has a negative impact in the model-based methods performance, favouring the usage of expert judgement. In practice, most organisations use expert judgment, making themselves dependent on the expert. A common problem found is that the performance of the expert’s estimation depends on his previous experience with identical projects. This means that when new types of projects arrive, the estimation will have an unpredictable accuracy. Moreover, different experts will make different estimates, based on their individual experience. As a result, the company will not directly attain a continuous growing knowledge about how the estimate should be carried. Estimation models depend on the input information collected from previous projects, the size of the project database and the resources available. Altran currently does not store the input information from previous projects in a systematic way. It has a small project database and a team of experts. Our work is targeted to companies that operate in similar contexts. We start by gathering information from the organisation in order to identify which estimation approaches can be applied considering the organization’s context. A gap analysis is used to understand what type of information the company would have to collect so that other approaches would become available. Based on our assessment, in our opinion, expert judgment is the most adequate approach for Altran Portugal, in the current context. We analysed past development and evolution projects from Altran Portugal and assessed their estimates. This resulted in the identification of common estimation deviations, errors, and patterns, which lead to the proposal of metrics to help estimators produce estimates leveraging past projects quantitative and qualitative information in a convenient way. This dissertation aims to contribute to more realistic estimates, by identifying shortcomings in the current estimation process and supporting the self-improvement of the process, by gathering as much relevant information as possible from each finished project.

Component-based software engineering: a quantitative approach

Dissertação apresentada para a obtenção do Grau de Doutor em Informática pela Universidade Nova de Lisboa, Faculdade de Ciências e Tecnologia

Derivation and consistency checking of models in early software product line engineering

Dissertação para obtenção do Grau de Doutor em Engenharia Informática

Information technology service management: an experimental approach towards IT service prediction

Dissertation presented to obtain a Masters degree in Computer Science

Software languages engineering: experimental evaluation

Dissertação apresentada na Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para obtenção do grau de Mestre em Engenharia Informática

A semi-automatic approach to code smells detection

Eradication of code smells is often pointed out as a way to improve readability, extensibility and design in existing software. However, code smell detection remains time consuming and error-prone, partly due to the inherent subjectivity of the detection processes presently available. In view of mitigating the subjectivity problem, this dissertation presents a tool that automates a technique for the detection and assessment of code smells in Java source code, developed as an Eclipse plugin. The technique is based upon a Binary Logistic Regression model that uses complexity metrics as independent variables and is calibrated by expert‟s knowledge. An overview of the technique is provided, the tool is described and validated by an example case study.

Evaluation of uv spectrophotometry for estimation of nitrite and nitrate in nitrified urine

Water is a limited resource for which demand is growing. Contaminated water from inadequate wastewater treatment provides one of the greatest health challenges as it restricts development and increases poverty in emerging and developing countries. Therefore, the connection between wastewater and human health is linked to access to sanitation and to human waste disposal. Adequate sanitation is expected to create a barrier between disposed human excreta and sources of drinking water. Different approaches to wastewater management are required for different geographical regions and different stages of economic governance depending on the capacity to manage wastewater. Effective wastewater management can contribute to overcome the challenges of water scarcity. Separate collection of human urine at its source is one promising approach that strongly reduces the economic and load demands on wastewater treatment plants (WWTP). Treatment of source-separated urine appears as a sanitation system that is affordable, produces a valuable fertiliser, reduces pollution of water resources and promotes health. However, the technical realisation of urine separation still faces challenges. Biological hydrolysis of urea causes a strong increase of ammonia and pH. Under these conditions ammonia volatilises which can cause odour problems and significant nitrogen losses. The above problems can be avoided by urine stabilisation. Biological nitrification is a suitable process for stabilisation of urine. Urine is a highly concentrated nutrient solution which can lead to strong inhibition effects during bacterial nitrification. This can further lead to process instabilities. The major cause of instability is accumulation of the inhibitory intermediate compound nitrite, which could lead to process breakdown. Enhanced on-line nitrite monitoring can be applied in biological source-separated urine nitrification reactors as a sustainable and efficient way to improve the reactor performance, avoiding reactor failures and eventual loss of biological activity. Spectrophotometry appears as a promising candidate for the development and application of on-line nitrite monitoring. Spectroscopic methods together with chemometrics are presented in this work as a powerful tool for estimation of nitrite concentrations. Principal component regression (PCR) is applied for the estimation of nitrite concentrations using an immersible UV sensor and off-line spectra acquisition. The effect of particles and the effect of saturation, respectively, on the UV absorbance spectra are investigated. The analysis allows to conclude that (i) saturation has a substantial effect on nitrite estimation; (ii) particles appear to have less impact on nitrite estimation. In addition, improper mixing together with instabilities in the urine nitrification process appears to significantly reduce the performance of the estimation model.

A debugging engine for parallel and distributed programs

Dissertação apresentada para a obtenção do Grau de Doutor em Informática pela Universidade Nova de Lisboa, Faculdade de Ciências e Tecnologia.

GVSOS: A New Client for OGC SOS Interface Standard

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Adding SOS Core & transactional profiles support to a web thin client

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Estudo técnico-económico de um sistema solar termoelétrico com back-up a biomassa

Este trabalho consistiu no estudo técnico-económico (dimensionamento) de uma central térmica solar de torre norte, com back-up a biomassa (pellets) para produção de energia elétrica. A partir da plataforma PVGIS obteve-se a irradiação normal direta, DNI, da região do Alentejo, Portugal, segundo critérios descritos na secção 3.1. O valor anual obtido foi de 2319 !"ℎ/%&, valor este que foi convertido em energia elétrica, obtendo-se, assim, um montante de 105.761.800 kWh por ano, em 25 anos. Estas importâncias foram obtidas através dos equipamentos que constituem a central, entre eles o campo solar (composto por 2067 helióstatos com área total de 188097 %&), a torre (convertendo a concentração da DNI para energia térmica ao fluído de transferência de calor, HTF), o bloco de potência (composta por permutador de calor, turbina, gerador elétrico e outros) e com ajuda de uma caldeira a biomassa (pellets) que consome cerca de 24.001,23 toneladas de pellets por ano. O estudo económico foi realizado num modelo executável no Excel, calculando, assim, o custo total do sistema, 138.515.886,87 €. A partir deste valor calcularam-se os indicadores económicos (Valor Atualizado Líquido, VAL=-12.282540,28 €) com uma tarifa bonificada de 0,22114 €/kWh, que indica que o sistema termo solar com back-up a biomassa (pellets) não é viável economicamente (e o mesmo acontece com o sistema termo solar sem a biomassa). O dimensionamento técnico do sistema CSP de torre é feito a partir de um modelo matemático executado em Excel e MatLab e pelo software System Advisor Model, SAM e o da caldeira a pellets é concretizado também no Excel. Os resultados do modelo matemático foram comparados com os valores obtidos no SAM para o mês típico de julho comprovando, assim, as fórmulas e conceitos contidos no estado da arte. Realizou-se, também, uma comparação dos valores anuais do sistema obtidos no SAM com os valores reais da Gemasolar (em Espanha), demonstrando então a veracidade dos valores obtidos no dimensionamento feito para Portugal.

Semantic architecture for sensors

Based on the report for the unit “Sociology of New Information Technologies” of the Master on Computer Sciences at FCT/University Nova Lisbon in 2015-16. The responsible of this curricular unit is Prof. António Moniz