In order to avoid the emission occurring due to the traditional ICE based Vehicles the majority of the population is shifting towards EVs as they have the capability to decrease the emission by 45 percent, but with EVs, there is an issue of range anxiety due to the limited power supply in existing charging infrastructure which prevents the charging of many EV simultaneously, In order to resolve this smart charging algorithm which utilizes Machine Learning (ML) to learn the EV charging behavior pattern, In order to maximize the equal share among EVs by prioritizing for an equal chance to reach a sufficient state of charge by the end of charging along with this precise charging time can mitigate the problem caused to drivers due to unavoidable charging behavior, in order to address these issue there is a need for accurate prediction of session duration or departure time. In, this research we have utilized the ACN dataset for historical charging data. Along with this, we have also utilized real-time climatic conditions in order to get more accurate and realistic results of the prediction by using various ML and Neural Networks out of which we got the best results through Random forest(RF) and Ensemble model with MAE of 30.75 minutes and 19.97 minutes, the prediction result revealed that addition of climatic data helps in improving the prediction accuracy and provide better results.