First, the client wants to know whether it is possible to use machine learning to identify the topic of a petition, for some specific topics of interest. Specifically, they want to know whether a given piece of text is about "education", "uk government and devolution", "health and social care", "london", "economy, labour and welfare", "environment and animal welfare", or "culture, sport and media" (7 classes in total). They also provide a feature indicating whether each petitions contains references to an event, a date, and/or a person, which they think might provide further relevant information. a) For the avoidance of doubt, the topic to predict is in the column: "petition_topic". The input features to use are: "petition_text" and "has_entity". b) The client will consider the results successful if 4 conditions are met: 1) The test accuracy is at least 86%; 2) The comparison of training and test accuracy does not suggest overfitting; 3) For at least 5 of the 7 classes, no more than 13% of the petitions in each of those classes get misclassified into an unrelated class. They want to know which classes meet this threshold. 4) And if petitions predicted as belonging to the "uk government and devolution