In this study, we propose a novel application of the Support Vector Regression (SVR) method to model a task variable in the Task-Technology Fit (TTF) theory. The support vector approach learns a parsimonious regression model from the given data to avoid the data over-fitting problem. Founded on the theories of statistical learning, mathematical programming and functional analysis, SVR is shown to outperform the traditional multiple linear regression method from the perspective of regression accuracy. Using a bootstrap procedure, we design a mechanism to extract significant factors from the support vector approach.