Modelling of Random Textured Tandem Silicon Solar Cells Characteristics: Decision Tree Approach

Abstract

We report decision tree (DT) modeling of randomly textured tandem silicon solar cells characteristics. The photovoltaic modules of silicon-based solar cells are extremely popular due to their high efficiency and longer lifetime. Decision tree model is one of the most common data mining models can be used for predictive analytics. The reported investigation depicts optimum decision tree architecture achieved by tuning parameters such as Min split, Min bucket, Max depth and Complexity. DT model, thus derived is easy to understand and entails recursive partitioning approach implemented in the “rpart” package. Moreover the performance of the model is evaluated with reference Mean Square Error (MSE) estimate of error rate. The modeling of the random textured silicon solar cells reveals strong correlation of efficiency with “Fill factor” and “thickness of a-Si layer”