Global temperature changes impact on crop production thus illuminating the highly sensitive nature of food security in the global landscape. Utilizing advanced crime models and deep learning solutions for the predictions of crop yields and plant diseases with the help of climatological conditions. The assessment is done using techniques such as Random Forest, Naive Bayes, Extra Tree, VGG16, VGG19, ResNet 101, and the results yielded confirm high levels of correctness and prediction abilities. The methodology integrates data from three distinct datasets: One is for the data on climatic records and soils to be used for crop categorization, the second is for the estimation of crop yield by regression analysis and the third for the classification of images to detect plant leaf diseases. In preprocessing, first the dataset is cleaned, followed by label encoding for categorizing it next normalization is performed and finally data augmentation is provided for handling class imbalance. Specific insights state that for Naive Bayes the highest classification accuracy is estimated to be 99%. 54%, The added features of Extra Trees show the best regression results in terms of R-squared being 0. 99 and Root-Mean-Square Error, RMSE of 8225. In regression scenario, ResNet18 reaches a model correlation coefficient of 93, while ResNet101 reaches a test accuracy of 96% in the image classification task. This paper has presented the need for data pre-processing as well as applying various algorithms and approaches in improving agricultural productivity following concern around climate change. Work that remains for the future includes using the existing models to improve agriculture in several environments, analyzing datasets for a more extensive range of agricultural conditions and incorporating more real-time data to implement models that would lead to the higher levels of sustainable and resilient agricultural practices.