Irrigation is crucial to agriculture. For available smart irrigations, accuracy is still a concern. Also, many of these systems are not affordable with practical usability, especially in rural areas of Maharashtra state. The proposed approach performs smart irrigation using Internet of Things (IoT) sensors, which collect the data about the water level in the agricultural field and provide the details of weather conditions, also uses Genetic Algorithm to optimize the input data, and Linear Regression model is used to predict the necessary water demands for the agricultural field. Here, the combination of Genetic Algorithm and linear regression is the useful and novel combination which helps to generate efficient results. By considering the level of soil moisture available in the crop field, the linear regression based model predicts the water needs and accord-ingly drives the motor pump for watering.  The accuracy is found to be better as compared to existing approaches with 98.33% and the Mean Squared Error (MSE) is observed as 0.017, which is very less. The water demand predictions are given for various value ranges of soil moisture and temperature. For 278.93wfv soil moisture and 36°C temperature, the predicted water demand by model is 6.11mm against the actual water demand of 6.28mm. The finding justifies the usability of the system in real-time agricultural fields as given in Results section.