Productivity of any crop through its life cycle of agriculture depends on various ground-weather indicators vitally, root level moisture, ground level moisture at 5 cm – 2 m. Estimation of soil moisture enhances the crop yield with right amount of watering process. This research work proposes prediction of ground level soil moisture through regression based machine learning models: linear, multi-liner, polynomial, random forest and exponential smoothing. The performance are assessed by Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE) and R2 metric. Aiming the reliable prediction, this work build on the real-time weather-related observations of the city ‘Madurai’ presenting a day observation as sequence of four observations that recorded at every six hour interval and data of ten year period from 2011 to 2022 has been used for training purpose; has shown minimum 8% improvement over existing methods when assessed in R2 metric.