The IT service business provides service request (SR) support to sales who are negotiating with customers on all aspects of sales (product/proposal configuration, contracts, billing, and invoicing). In order to manage appropriate staffing levels, one needs to know the volume of demand for SR in ahead. This forecasting work is challenging because of the hundreds of SR features and the huge amount of SR with large varieties. However, it is the support team‚Äòs responsibility to respond to each SR in time with high quality. To forecast the SR volume, advanced techniques of pre-processing data and human- in-the-loop were developed and well applied in this project.