Dept-of-Comp-Sci-University-of-Ghana/team-task-1---natural-language-with-disaster-tweets-cmsm-gp-9

team-task-1---natural-language-with-disaster-tweets-cmsm-gp-9 created by GitHub Classroom # Twitter Disaster Detection - Model Evaluation Report ## Introduction The objective of this analysis is to build a machine learning model that can accurately predict whether tweets are about real disasters or not. Accurate detection of disaster-related tweets can help in identifying and responding to real-time emergencies and improve crisis management. In this report, we will present the results of our analysis, including the dataset description, model evaluation metrics, and a summary of the findings. ## Dataset Description The dataset used for this analysis consists of 10,000 hand-classified tweets. The training dataset has the following shape: (7613, 5), and consists of the following columns: `id`, `keyword`, `location`, `text`, and `target`. The `target` column represents the binary label indicating whether the tweet is about a real disaster (1) or not (0). The class distribution in the training data is as follows: Non-Disaster Tweets (Class 0): 4342, and Disaster Tweets (Class 1): 3271. The test dataset has the shape (3263, 4) and contains columns: `id`, `keyword`, `location`, and `text`. ## Model Evaluation Results We evaluated three different models using the TF-IDF vectorization technique. The models used were Logistic Regression, Naive Bayes, and Support Vector Classifier (SVC). Here are the results of the model evaluation: ### Logistic Regression Model - Accuracy: 0.7806479859894921 - F1 Score: 0.7364544976328249 - Precision: 0.7486631016042781 - Recall: 0.7246376811594203 ### Naive Bayes Model - Accuracy: 0.8012259194395797 - F1 Score: 0.7440811724915444 - Precision: 0.8168316831683168 - Recall: 0.6832298136645962 ### Support Vector Classifier (SVC) Model - Accuracy: 0.8016637478108581 - F1 Score: 0.7367809413131902 - Precision: 0.8397350993377484 - Recall: 0.6563146997929606 The logistic regression and SVC models achieved similar accuracies of approximately 80%, while the Naive Bayes model performed slightly better with an accuracy of 80.12%. T …