Watering to the plants in the agriculture field is done promptly in conventional farming, without considering important parameters such as the water requirement of a crop, the possibility of rain on the next day, etc. In some instances, soil moisture, humidity, and temperature are taken into consideration for crop watering which may not be sufficient to conserve the water. The atmospheric conditions play a main role in the water requirements of the crop. The internet of things and machine learning are extensively becoming popular during the last decade and find applications in all domains such as agriculture, banking, smart home, etc. Internet of things is used in this agriculture application, where the data from soil moisture sensor, humidity, and temperature sensor is combined along with weather data to water the crops. In this research paper, an IoT-based smart watering system for assessing the watering requirements of the crop is proposed. The soil moisture parameters in multi-crop environments at various locations are measured using a ground moving robot with a moisture sensor embedded in it. Information gain and entropy statistics of the decision tree are applied to find the status of the output actuator (sprinkler motor). Entropy methods make our system more efficient by using decision tree split criteria, and it is implemented using MATLAB where the information gain and entropy are used for the selection of the best feature to split the tree. Accuracy rate and decision tree performance are improved by the efficient feature split of decision trees. An accuracy of 96.5 % is achieved for the proposed system. The proposed model is a low-cost system where a single ground moving robot is used to collect the crop parameters. The system will reduce the time a farmer spent in the agriculture field as he need not visit the farm regularly.