This paper presents smart IoT networks and machine learning algorithm to monitor the aquatic environment. The growth and expansion of invasive plants in freshwater lakes requires an accurate monitoring and long-term data collection tool, which the system proposed in this paper is addressing. The expansion of invasive weed water hyacinth highly affects the biodiversity and environment of freshwater lakes and rivers. Dynamic and continuous environmental monitoring system increases the prediction accuracy of machine learning algorithms by delivering real time data. Predicting the seasonal and daily water hyacinth weed expansion rate intensity of lake Tana improves the biodiversity of the lake and living standards of the community. Laboratory based experimental data is used to predict the relationship between the intensity of water hyacinth expansion rate and influencing environmental features such as depth, temperature, pH, total nitrogen, total phosphorous and chlorophyl-a contents of lake Tana. Four machine learning techniques namely multivariate linear regression, random forest, support vector machine and extreme gradient boosting algorithms are applied for data analysis. The random forest machine learning algorithm predicts water hyacinth annual expansion rate intensity better than others.