Estimation of Meterological Drought Prediction by SPEI Using SVM

Water is an important element found on Earth and in modern days scarcity of water leads to drought. Drought prediction is the process to predict the occurrence of drought with the use of several components such as wind speed, temperature, humidity, rainfall. This model is based on Machine Learning mainly using Support Vector Machine (SVM) Algorithm and SPEI data sets. Standardized evapotranspiration precipitation index (SPEI) proposed for drought analysis as a case study in Asian countries. We looked at two composite time periods ranging from three months (SPEI-3) to six months (SPEI-6), and we came to the conclusion that employing SPEI-3 would be more manageable and predictable than the established method. Priority over Random Forest (RF) algorithm due to increased accuracy. Identification of sub-regional patterns of drought co-variability using principal component analysis (PCA). From the original SPEI data, a few new representative ones were created using the PCA. Another algorithm that should be implemented is the gradient boosting machine algorithm. The approach is built on the concept of boosting, which incorporates all of a group of weak learners forecast to generate a strong learner using extra training methods. Our project's primary goal is to use SVM to make predictions with greater accuracy.