The Caesar cipher is one of the most basic cryptographic methods that was invented by Julius Caesar himself. This encryption method was used by him to safely communicate with soldiers. While simple, analysis and techniques to crack the cipher have been questions people have pondering since the time of the emperor. One such technique is letter frequency analysis, which uses the distribution of letters in English text to make predictions. In this paper, we will be exploring how letter frequency and machine learning regression models can be used to crack the Caesar cipher. We will be analyzing the performance of several models, optimize the best ones and create our final key guessing models. Our end result are multiple regression models that reliably predict the key of a Caesar cipher.