Abstract
Software Development Organizations (SDO) develop a massive number of projects per year. One of the elementary and significant features of any SDO is to use a tool that can precisely estimate the software cost. It directly affects nearly all management activities including resource allocation, project planning, and project bidding. Imprecise estimation causes troubles e.g. dropping the worth of the project, waste the company's budgets and can outcome in the disaster of the project. During the last few decades' researchers have developed a large number of models for software cost estimation (SCE). However, SCE is still a challenging task. Algorithmic and non-algorithmic approaches were firstly used to achieve the goal. Each of them has their own merits and demerits but still, these are considered as primary tools for SCE. This study proposes Flower Pollination Algorithm (FPA) for SCE. Mean Magnitude of Relative En or (MMRE) is used as an evaluation metric for benchmarking the proposed model with the existing model. All the results of FPA are compared with the COCOMO model. Experimental results show a better performance of FPA as compare to COCOMO. Three datasets from NASA software projects are selected, NASA93, NASA63, and NASA60. On NASA93 dataset the improvement is 10.17%, on NASA63 the improvement is 77.38% and on NASA60 it is 22.96%.