Objectives: One of the major challenges facing the cancer biology is finding the best treatment, (the most efficient and least side effects). Using microarray data led to fundamental changes in prediction clinical outcomes. Analyzing microarray data due to large number of variables in comparison with the number of samples needs for appropriate methods. The aim of the present study was to evaluate and to compare two data mining methods for classification of cancer patients. Methods: This study used two public dataset (lung and breast cancer) for classification of tumor types and other outcomes. We applied wavelet transform for feature extraction and neural network as a classifier. The accuracy criterion was used to evaluate artificial neural network performance. Results: Accuracy of artificial neural network in lung and breast cancer data was 100%. Dimension reduction did not change the accuracy for lung cancer dataset but it slightly declined for the breast cancer dataset. Conclusion: Artificial neural network was highly efficient in determining tumor type using microarray data compared with other classification methods. The results indicated that feature extraction and dimension reduction with wavelet transform did not change the accuracy of artificial neural network for data with large enough sample size.
