Economy of India profoundly relies upon farming. Conventional methods of suggestions are still utilized for farming. Farming is the cornerstone of emerging nation like India. For the income, most of their populace relies upon farming. As of now, farming is based on different approximations of fertilization amount and the sort of yield to be developed or planted. Agriculture holds a predominant position in the growth of any country's prosperity. However, there exists a major threat in the crop yield due to unpredictable and uncontrolled climatic changes, traditional farming methods, and poor irrigation services [10]. Machine learning is an arising research field in analysis of crop yield. Yield forecast is a very significant issue in agriculture. Machine learning methods are the better decision for this issue. Different Machine learning methods are utilized and assessed in agribusiness for assessing what's to come year's production of crop. This paper proposes the Improved Support Vector Machine (Tree-structured Parzen Estimator (TPEOSM)) is applied in the agricultural dataset to predict the crop yield with fertilizer amount. The performance metrics such as precision, recall, f1 score, and accuracy is evaluated. The proposed method accuracy is better than the traditional SVM classifier. Thus, we hope it will help the farmers to select the right crop for the cultivation in basis of temperature, rainfall and other weather conditions and amount of fertilizers thus to be used for the particular crop.