Accurate weather forecast is important to our daily life and have both economic and environment impact. Through physical atmospheric models, a short period time weather can be accurately forecasted. To provide weather forecast, machines learning techniques can be used for understanding and analyzing weather patterns. In this paper, we propose a deep learning-based weather forecast system and conduct data volume and recency analysis by utilizing a real-world weather data set as a case study to demonstrate the learning ability of deep learning model. By using the Python Keras library and Pandas library1, we implement the proposed system. Based on the system, we find out not only the relationship between the prediction accuracy and data volume, but also the relationship between the prediction accuracy and data recency. Through extensive evaluations, our results show that according to the weather data we have been using, more data is beneficial to increasing the accuracy of a trained model. The recency of the data does not have a consistently significant impact on the accuracy of the trained model.1Certain commercial equipment, instruments, or materials are identified in this paper in order to specify the experimental procedure adequately. Such identification is not intended to imply recommendation or endorsement by the National Institute of Standards and Technology, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose.