Web has performed a vital part in the advancement of modern technology, making it easy for the people to access information from anywhere in the world. A huge amount of data is available on the web that are both secured and unsecured. Secured data can only be accessed by individuals with proper authentication credentials, while unsecured data can be accessed easily. The data and services are accessed on the web via URL (Uniform Resource Locator). With the increasing frequency of secured data on the web, there has been an equivalent rise in the number of cyber attacks that are aimed in gaining the unauthorized access to the secured data via malicious URLs. But most of these attacks can be reduced if the URL is legitimate. By extracting and evaluating the different parts of the URL, it is easy to conclude whether the URL is malicious or legitimate. So, in order to identify and classify these URLs to be malicious or legitimate, various machine learning algorithms are available and one can employ deep learning strategies to implement this problem statement. The Highest obtained training accuracy is 99.02% and highest testing accuracy obtained is 97.06%.