The ascent prediction of high-altitude zero-pressure stratospheric balloons is an important aspect of targeted test flight. Prediction of the balloon ascent rate is the prerequisite for many of the flights as it helps in planning ballasting and valving manoeuvres. In this paper, a standard analytical model, a fuzzy model and a statistical regression model are developed and compared to predict the zero-pressure balloon ascent. The flight data is extracted from the Esrange balloon service system for zero-pressure balloons with different payload capability, and several potential explanatory variables are computed for every sampled climbed segment. For the fuzzy modelling approach, a fuzzy c-mean clustering algorithm is used for system identification and prediction. For the regression approach, a Gaussian process regression is used, and principal component analysis is applied for finding the significant inputs. The result shows that the data driven approaches are more efficient than the standard analytical model.