Predicting criminal activity is still a major obstacle to maintaining community safety. Even though statistical methods and historical data provide insightful analysis of crime, traditional crime analysis frequently fails to capture the complexity of contemporary crime patterns. This research presents a unique framework to improve crime prediction capabilities by utilizing time series analysis and machine learning. Our research aims to develop a multi-faceted approach with three distinct but interconnected objectives. Firstly, we will explore supervised learning algorithms to predict the most likely crime types based on historical data and relevant features like time of day, location type, and past crime occurrences. This will allow for proactive strategies by understanding the potential criminal activity. Secondly, spatial analysis techniques will be employed to identify high-risk zones with a high probability of future crime occurrences. We used the more than 6,000,000 records in the Chicago Police Department's CLEAR (Citizen Law Enforcement Analysis and Reporting) system to test our models. This will provide a more granular understanding compared to traditional methods, enabling focused deployment of resources and targeted prevention efforts. Finally, we will delve into time series forecasting models to predict future crime likelihood within the identified high-risk zones. This will provide valuable insights into potential seasonal variations or crime surges, allowing for better preparation and resource allocation. This research aims to create a complete framework for crime prediction and prevention tactics by merging several methodologies in order to address the shortcomings of existing approaches. All people's lives might be far safer and more secure with the help of this framework. Additionally, this work sheds light on the suitability of several machine learning models for the analysis of crime report datasets from sizable cities.