Crop yield prediction is a difficult task in agriculture since it depends on a variety of factors, including weather patterns, pest and disease activity, soil quality, and management techniques. Although conventional techniques for predicting agricultural production were originally thought to be adequate, the environment of today is much more complicated, and seasonal changes are no longer enough to explain variances in accuracy. Even the most seasoned specialists are challenged by the dizzying number of variables that climate change has introduced to the equation. The fine balance between sowing and reaping feels more fragile due to fluctuating weather patterns, severe occurrences, changed growth seasons, and changing pest pressures. In spite of this uncertainty, however, state-of-the-art instruments and quick thinking have led to the development of the novel yield sampling technique as a stopgap measure to raise crop yield forecast accuracy. In order to provide more accurate yield calculations, this paper suggests a machine learning based method that combines rainfall, fertilizer, and historical data. The raw data captured from different sources are preprocessed using feature scaling and feature selection techniques before used for training the machine learning algorithms. Predictive analytics based estimation has been proved as a robust technique in the field of data mining.