Predicting the Duration of User Stories in Agile Project Management

Effective effort estimation in agile project planning is vital because it helps organizations build product plans that they can stick to, have shorter turn-around time, and have better cost discipline. Machine learning can play an essential role in planning and estimating the project schedule. In this paper, a series of supervised machine learning models were studied, analyzed, and implemented to solve the problem of predicting effort estimates in Agile Scrum. The obtained results are compared with similar previous studies. We performed experiments using different Natural Language Processing (NLP) methods such as Term Frequency-inverse document frequency (TF-IDF), fastText, and different Neural Networks, including Recurrent Neural Networks (RNN), Long Short Term Memory (LSTM), and Bidirectional Encoder Representations from Transformers (BERT). The trained models were fitted with three publicly available datasets. Our findings show that fastText (with a pre-trained model) significantly performed better in predicting the story-points of user-stories. The second-best performing model was bidirectional LSTM. Moreover, distilBERT performs poorly among all the models analyzed. This study can pave the way for organizations to benefit from these machine learning models and accurately predict project deadlines and schedules.