Software development framework on small team using Agile Framework for Small Projects (AFSP) with neural network estimation

Technology development in growing companies is important because it has great potential to reduce cost, time of work, efficiency, flexibility. In small developer team, finishing many software development projects with tight schedule is a challenge to work efficiently and effectively. Wrong estimation could delay the other project completion and lower the team's assessment point. This study focuses on developing framework that fits well with small-scale developers by combining Agile Framework For Small Projects (AFSP) and Neural Network estimation. The research method is to identify 5 risks based on the agility factor of each project. Then determine the appropriate agile practices. Next, do the selection of dataset agile characteristics, and calculate the Neural Network estimation. Implementation is done with estimation result guidance. The project that has been done is assessed to measure the level of agility of project result. Then, the result of project's succession and accuracy are analyzed. This research is implemented on six projects in fast moving consumer goods (FMCG) company. The results showed all projects are agile, four projects with small Mean Relative Error (MRE) values and two projects with big MRE values.