Social mediaSocial media has been usedSri Lankan floods widely during crises and disasters. The data generated by social mediaSocial media is huge and has the potential for deep analysis. The aim of the study is to find out the social mediaSocial media uses in different phases of disaster situations like pre-, during-, and post-. The study will help in understanding the categories for which social media in particular Twitter is used and also to find out the areas in which attention is needed for further effective communicationCommunication. ML (Machine Learning) algorithms like Decision Tree, Naive Bayes, and IBK are implemented on the categorised data and the classes are predicted by the algorithms. The Twitter data was collected during Sri Lankan FloodsSri Lankan floods and CycloneCyclone OckhiCyclone Ockhi which occurred during the year 2017. The collected data is then categorised using Keyword analysis and then the data is analysed for trends of usage. Also, this data is given as input to an open-source visualisation tool called WEKA. It was found that social mediaSocial media had political factors influencing the tweet activity during Cyclone OckhiCyclone Ockhi. The study also identifies the opportunities and challenges faced in disaster situations using social mediaSocial media like information about missing people which is underutilised by the line departments of government and the potential of social mediaSocial media in disseminating search and rescue measures. The study also tries to list out the critical impacts of social media on the user communityCommunity.