Apaulgithub/oibsip_taskno4 ? reverse-engineered prompt

Build me a complete email spam detection notebook using the spam.csv dataset. I want it to clean and prepare the text data, explore the dataset a bit, and show a few simple visuals like spam versus ham counts and common words in spam messages, including a word cloud if it makes sense.

Then train a few machine learning models for classifying messages as spam or not spam, and compare them clearly so I can see which one works best. Please include the usual evaluation results like accuracy, precision, recall, F1 score, and a confusion matrix. If one model stands out, make that the main one and show how to use it on a new message.

Keep it beginner friendly and easy to run in Jupyter or Colab. If you need to look up current best practices for text preprocessing or plotting, feel free to do that online. Make the notebook neat, readable, and complete enough that someone could open it and understand the whole process from start to finish.