Predicting diffusion of innovative products using neural networks

Predicting market growths of innovative products are essential for policy makers, market planners and various hardware and software companies. However, it is difficult to find a model that generalises because both internal and external factors influence the growth process. This study investigated models based on diffusion and connectionist theories to predict diffusions of innovative products. This paper shows that a simple Multi‐Layered Perceptron (MLP) neural network can create a very flexible response function to forecast generic diffusion patterns of innovation processes. This study compared performances of MLP and diffusion models on simulated data with varying degrees of uncertainties. MLP models outperformed diffusion models.