An Approach to Probabilistic Effort Estimation for Military Avionics Software Maintenance by Considering Structural Characteristics

Stochastic gradient boosting for predicting the maintenance effort of software‐intensive systems

“…Regarding SMEP, we only identified two studies in the last five years [13,14], 13 studies between 2001 and 2013 [12,[15][16][17][18][19][20][21][22][23][24][25][26], and one of 1995 [27]. The models applied in these 16 studies were statistical regressions [12-18, 21-23, 25-27], neural networks [13,15,[17][18][19]27], SVR [14], rule-based systems [14,15], DT [14], Bayesian network [20], analogy [24], and a pattern recognition approach termed optimised set reduction [27]. In one of the two studies that use the same datasets than our reports that there was no statistically significant difference among the accuracies of a SLR and three types of neural networks [13] when a data set was used, whereas the second one reports that the type of ɛ-SVR was statistically better than SLR, three types of neural networks, ARu, and DT, for one of the five data sets used [14].…”

“…In spite of SM corresponds to the longest phase of the software life cycle and, in most cases, the most expensive [8], SM has not received the same degree of attention that the other phases have [3]. Regarding SMEP, we only identified two studies in the last five years [13, 14], 13 studies between 2001 and 2013 [12, 15–26], and one of 1995 [27]. The models applied in these 16 studies were statistical regressions [12–18, 21–23, 25–27], neural networks [13, 15, 17–19, 27], SVR [14], rule‐based systems [14, 15], DT [14], Bayesian network [20], analogy [24], and a pattern recognition approach termed optimised set reduction [27].…”

“…Based upon SM consumes a major share of the financial resources in a software life cycle, SMEP is an important aspect in the planning phase [3]; however, to date, we only identified 16 SMEP studies published since 1995 [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27].…”

Estimativa de Esforço em Atividades de Manutenção de Software: Um Mapeamento Sistemático

“…Pontos de Função e Analogia [Leung 2002] C2 Analogia com o método vizinho virtual (AVN) [Ahn et al 2003] C1 Pontos de Função [Hayes et al 2004] C1 Modelo adaptável baseado em métricas [De Lucia et al 2005] C1, C2 Regressão linear multivariada [Song et al 2007] C1, C2 Modelo probabilístico baseado em rede Bayesiana [Shukla and Misra 2008] C1, C2 Rede Neural [Tenório Jr et al 2008 [Ahn et al 2003] Pontos de Função [Hayes et al 2004] Linhas e operadores alterados [De Lucia et al 2005] Tamanho do sistema (KLOC), número de tarefas [Song et al 2007] Experiência do mantenedor, tamanho do software, características estruturais [Shukla and Misra 2008] Complexidade, número de linhas, número de arquivos [Tenório Jr et al 2008] Pontos de Função [Nguyen et al 2011] Linhas inseridas, modificadas e excluídas [Alomari et al 2014] Tamanho total do sistema, tempo de atraso, intervalo de tempo entre versões, número de hashes modificado [Chandra et al 2017] se tornando o segundo método mais aplicado. Dos demais métodos podemos citar: julgamento de especialista na atividade, similaridade textual e analogia que fazem a busca de atividades já realizadas para assim gerar a estimativa de esforço, dentre outros métodos apresentados na Tabela 1.…”

Towards apriori and posteriori estimation models for renewable energy projects in software engineering

“…On the other hand, there are many other estimating techniques and models constructed and developed from information available after the completion of projects [3] such models should be qualified as a posteriori estimation models, such as COCOMO [4], SLIM [5], Checkpoint [6] , PRICE-S [7], SEER [8], Walston-Felix Model [9], BaileyBasili Model [10], Boeing Model [11], Doty Model for KLOC [12], Albrecht and Gaffney Model [13], Kemmerer Model [14], Matson, Barnett & Mellichamp Model [15].…”