Flood prediction based on Machine learning techniques is proposed in our process. Machine learning methods were supervised methods for the classification of the data (i.e., labels has to be provided for each training data that is used). The flood prediction system helps in real time monitoring and flood forecasting process. The input data like Water level, Capacity, Inflow, Discharge rate, Evaporation loss and rainfall were collected in particular duration of time for a particular Dam. The data were collected for 10 years and the values were calculated based on the different type of sensors for each type of data. The data itself is nonlinear and non-stationary. Unanticipated environmental conditions and limitations in the sensing and communications hardware cause the data to be corrupted by missing observations, spikes and multiple discontinuities. The probabilistic nature of the method is ideal for reporting uncertainty estimates to human operators. The approach can also be applied to detect patterns, other than faults, which are of great environmental significance. The calculated data were used for the prediction process. The input data were initially preprocessed by removing unnecessary informations like date, duration from the input data. The missing informations in the remaining data were removed by identifying the gaps in the data. The missing values were removed. The data is then optimized using Genetic algorithm for the selection of the best data from the input data. The selected data is then used for the classification process. The classification process is done based on kNN classifier. Different levels of warning were issued by the identification of different water level in the dam. The performance of the process is measured by the calculation of the performance metrics like accuracy, sensitivity and specificity of the classifier. The proposed method is proved to produce more similar predicted value compared to manual prediction.