In India, selecting the right crop is essential due to diverse Agro-climatic conditions, such as soil types and socio-economic factors. Traditional crop recommendation methods have evolved and advanced by utilising the recent technologies in Artificial Intelligence (AI), Machine Learning (ML), the Internet of Things (IoT), and computer vision, which have enabled smart crop recommendation systems based on soil and climatic conditions. This review compares the methodologies, input features, performance evaluation metrics, limitations, and practical applicability by examining the publicly available agricultural datasets, including soil-characteristics, environmental factors, and previously harvested crops available in the open agricultural datasets from sources like Open Government Data Platform India (n.d.) and Kaggle. This study identifies key research limitations, including the drawbacks of model generalisation or integrating varied data, explainability, scalability, and real-world applications in the agricultural field. By this structured literature, best performing model was identified on classification of the standard dataset into classes as rice, maize, chickpea, kidney beans, pigeon peas, moth beans, mung bean, black gram, lentil, pomegranate, banana, mango, grapes, watermelon, muskmelon, apple, orange, papaya, coconut, cotton, jute and coffee, by comparing the state of art models viz. SVM, K-Nearest Neighbours, Random Forest, Gaussian Naive Bayes, XGBoost, Gradient Boosting, AdaBoost, LightGBM, Bagging, and CatBoost. These models were evaluated on the following performance metrics as precision, recall, F1-score, accuracy and model inference time amongst which Gaussian Naive Bayes results to be accurately precise and fastest, making it effective for real-time applications.