Nowadays social media platforms such as Twitter provide a great opportunity to understand public opinion of climate change compared to traditional survey methods. In this paper, we constructed a massive climate change Twitter dataset and conducted comprehensive analysis using machine learning. By conducting topic modeling and natural language processing, we show the relationship between the number of tweets about climate change and major climate events; the common topics people discuss climate change; and the trend of sentiment. Our dataset was published on Kaggle (\url{https://www.kaggle.com/leonshangguan/climate-change-tweets-ids-until-aug-2021}) and can be used in further research.