In-season yield and commercial cane sugar (CCS) level estimation for sugarcane is critical for fine tuning the irrigation, fertilizer and pesticide schedules during the growing season to increase the yield and CCS and for optimizing the supply-demand logistics. In this paper, we present a case study of estimating in-season, field-level yield and CCS for sugarcane using crop attributes, weather and soil data as features for three corporate owned sugarcane farms in Thailand. We provide a detailed analysis of feature correlation, and accuracy of the yield and CCS estimation models across growing conditions and stages. Using various machine learning techniques to estimate yield and CCS, we obtained minimum normalized root mean square error (NRMSE) with linear regression. The values of minimum NRMSE for yield and CCS estimation are 0.197 and 0.203 respectively.