Web Engineering - Web Effort Estimation
Download as PPTX, PDF3 likes4,015 views
The document discusses various techniques for estimating effort in web development projects, including expert opinion, algorithmic models, and artificial intelligence. It describes expert opinion as a subjective approach based on previous experience, while algorithmic models attempt to formalize relationships between effort and factors like project size through equations. The document also explains case-based reasoning and regression trees as popular artificial intelligence techniques, with case-based reasoning estimating effort by comparing new projects to similar past projects.
Web Engineering - Web Effort Estimation
- 1. Web Effort Estimation & Effort Estimation Techniques By Nosheen Qamar University of Lahore, Pakistan
- 2. In Web Development, effort estimation is the process of predicting the most realistic amount of effort (expressed in terms of person-hours or money) required to develop or maintain software based on incomplete, uncertain and noisy input.
- 3. Software effort models and effort estimates help project managers allocate resources, control costs, and schedule and improve current practices, leading to projects that are finished on time and within budget. In the context of Web development and maintenance, these issues are also crucial, and very challenging, given that Web projects have short schedules and a highly fluidic scope.
- 4. The Web is used as a delivery platform for numerous types of Web applications, ranging from complex e-commerce solutions with back-end databases to on-line personal static Web pages. With the sheer diversity of Web application types and technologies employed, in order to make the Project successful, we need to estimate cost, effort, budget, resources, schedule etc.
- 5. Prediction is a necessary part of an effective process, whether it be authoring, design, testing, or Web development as a whole. A prediction process involves: ◦ The identification of measures (e.g. number of new Web pages, number of new images) that are believed to influence the effort required to develop a new Web application. ◦ The formulation of theories about the relationship between the selected measures and effort (e.g. the greater the number of new static Web pages, the greater the development effort for a new application). ◦ The capturing of historical data (e.g. size and actual effort) about past Web projects or even past development phases within the same project. ◦ The use of this historical data to develop effort estimation models for use in predicting effort for new Web projects. ◦ The assessment of how effective those effort estimation models are, i.e. the assessment of their prediction accuracy.
- 6. The purpose of estimating effort is to predict the amount of effort to accomplish a given task, based on knowledge of other project characteristics that are believed to be related to effort. Project characteristics (independent variables) are the input, and effort (dependent variable) is the output we wish to predict (see Fig. 2.1). For example, a Web company may find that to predict the effort necessary to implement a new Web application, it will require the following input: estimated number of new Web pages, total number of developers who will help develop the new Web application, developers’ average number of years of experience with the development tools employed, and the number of functions/features (e.g. shopping cart) to be offered by the new Web application.
- 8. Several techniques for effort estimation have been proposed in software engineering. These fall into three general categories ◦ Expert opinion ◦ Algorithmic models ◦ Artificial intelligence
- 9. Expert opinion represents the process of estimating effort by subjective means, and is often based on previous experience from developing/managing similar projects. It has been and still is widely used in software and Web development. The drawback of this technique is that it is very difficult to quantify and to determine those factors that have been used to derive an estimate. However, studies show that this technique can be an effective estimating tool when used in combination with other less subjective techniques (e.g. algorithmic models)
- 10. In terms of the diagram presented in Fig. 2.1, the sequence occurs as follows: ◦ An expert looks at the estimated size and cost drivers related to a new project for which effort needs to be estimated. ◦ Based on the data obtained in a) (s)he remembers or retrieves data on past finished projects for which actual effort is known. ◦ Based on the data from a) and b) (s)he subjectively estimates effort for the new project. Deriving an accurate effort estimate is more likely to occur when there are completed projects similar to the one having its Estimated size and cost drivers Deriving an effort estimate Estimated effort + effort accuracy Data on finished projects. Step 2 Step 3 Step 1
- 11. To date, the most popular techniques described in the effort estimation literature are algorithmic techniques. Such techniques attempt to formalize the relationship between effort and one or more project characteristics. The result is an algorithmic model. The central project characteristic used in such a model is usually taken to be some notion of software size (e.g. the number of lines of source code, number of Web pages, number of links). This formalization is often translated as an equation such as that shown by Eq. 2.1, where a and b are parameters that also need to be estimated. Equation 2.1 shows that size is the main factor contributing to effort, and can be adjusted according to an Effort Adjustment Factor (EAF), calculated from cost drivers (e.g. developers, experience, tools). An example of an algorithmic model that uses Eq. 2.1 is the COnstructive COst MOdel (COCOMO) model, where parameters a and b are based on the type of project under construction, and the EAF is based on 15 cost drivers that are calculated and then summed where: a, b are parameters chosen based on certain criteria, such as the type of software project being developed. EstSizeNewproj is the estimated size for the new project. EAF is the Effort Adjustment Factor.
- 12. The COCOMO model is an example of a generic algorithmic model, believed to be applicable to any type of software project, with suitable calibration or adjustment to local circumstances. In terms of the diagram presented in Fig. 2.1, the model uses parameter values that are based on past project data; however, for anyone wishing to use this model, the steps to use are 1, 2, and 3. Step 1 is used only once to calculate the initial values for its parameters, which are then fixed from that point onwards. The single use of step 1 makes this model a generic algorithmic model.
- 13. Regression-based algorithmic models are most suitable to local circumstances such as “in-house” analysis as they are derived from past data that often represents projects from the company itself. Regression analysis, used to generate regression-based algorithmic models, provides a procedure for determining the “best” straight-line fit to a set of project data that represents the relationship between effort (the response or dependent variable) and project characteristics (e.g. size, experience, tools, the predictor or independent variables). Regarding the regression analysis itself, two of the most widely used techniques are multiple regression (MR) and stepwise regression (SWR). The difference between both is that MR obtains a regression line using all the independent variables at the same time, whereas SWR is a technique that examines different combinations of independent variables, looking for the best grouping to explain the greatest amount of variation in effort. The sequence followed here is as follows: ◦ Past data is used to generate a cost model. ◦ This model then receives, as input, values for the new project characteristics. ◦ The model generates estimated effort. The sequence described here in corresponds to steps 1, 2, and 3 from Fig. 2.1
- 14. Artificial intelligence techniques have, in the last decade, been used as a complement to, or as an alternative to, the previous two categories. Examples include ◦ Fuzzy logic ◦ Regression trees ◦ Neural networks ◦ Case-based reasoning We will cover case-based reasoning (CBR) and regression trees (CART) as they are currently the most popular machine learning techniques employed for Web cost estimation.
- 15. Case-based reasoning (CBR) provides estimates by comparing the current problem to be estimated against a library of historical information from completed projects with a known effort (case base). It involves ◦ Characterizing a new project p, for which an estimate is required, with attributes (features) common to those completed projects stored in the case base. In terms of software cost estimation, features represent size measures and cost drivers which have a bearing on effort. ◦ Use of this characterization as a basis for finding similar (analogous) completed projects, for which effort is known. This process can be achieved by measuring the “distance” between two projects, based on the values of the number of features (k) for these projects. This is the common choice in Web and software engineering.
- 16. When using CBR there are six parameters to consider: ◦ Feature Subset Selection determining the optimum subset of features that yield the most accurate estimation. ◦ Similarity Measure measures the level of similarity between different cases, with several similarity measures proposed in the literature. ◦ Scaling Also known as standardization, represents the transformation of attribute values according to a defined rule, such that all attributes present values within the same range and hence have the same degree of influence on the results. ◦ Number of Analogies refers to the number of most similar cases that will be used to generate the estimation. ◦ Analogy Adaptation Once the similar cases have been selected the next step is to decide how to generate the estimation for project P(new). ◦ Adaptation Rules are used to adapt estimated effort, according to a given criterion, such that it reflects the characteristics of the target project more closely. For example, in the context of effort prediction, the estimated effort to develop an application would be adapted such that it would also take into consideration the application’s size values.