This paper is about developing machine learning models for identifying severity of Chronic Obstructive Pulmonary Disease (COPD) in patients. The motive is to apply machine learning techniques in identification of severity of COPD categorizing it into one of four stages which have been discussed briefly in the introduction part. By using a dataset of patients with COPD which included the patient data as well as severity of the disease, a machine learning model was developed by using supervised learning. The developed models demonstrated good cross validation accuracy. The first model developed using random forest algorithm showed accuracy of 86% which was then improved to 90% using the XGBoost algorithm. The dataset employed in this study is, however, small with only 101 samples and a larger dataset could potentially help improve the efficacy of the ML models. The outcome of the paper is the potential of ML models in identifying the severity of COPD based on spirometry data. After this study, we suggest that the applying of machine learning models along with regular clinical diagnosis can enhance the accuracy of the diagnosis. Further work in this area can help develop machine learning models which can help in increasing the accuracy of diagnosis of severity of COPD in patients.