Nowadays thyroid disease has become very common but it is important to check it at the right time otherwise it can turn into some serious disease. It is easy to do a blood test to check thyroid. But it can also contain some errors like noisy data. Therefore, to find out the exact disease related to thyroid, deep learning algorithm should be used. In today's modern era, special contribution of deep learning algorithms is made to process vast data. There are some deep learning tools that can process large datasets accurately and efficiently. Some such attempts are made to optimize classification models by using deep learning tools with small datasets. In this paper thyroid dataset from Kaggle is executed on data mining tool WEKA and deep learning tool TensorFlow, Keras. WEKA gives an accuracy of 94.77% and TensorFlow 92.70%. Our objective is to experiment with both the tools and find a better tool for future research.