We observe fire hazards causing life loss and property loss frequently in domestic and industrial scenarios. In industries we usually have many blocks or buildings, and it is impossible to check every building every second of the day. So, the authors' model continuously checks for fire and gives a signal: either buzzer or light depending on the requirement. This is an embedded way of hardware application and software. They also used different machine learning models and algorithms to predict the future time of the fire, using regression. For prototype applications, they use linear regression, and for real-time applications, they use k-means clustering or any other model for better accuracy.