Agriculture depends on many factors such as soil nutrients, temperature, humidity, rainfall, soil type and crop characteristics. Selecting a suitable crop and applying the right fertilizer are important for obtaining good production. Traditionally, farmers mainly depend on previous experience, local knowledge and manual soil testing. These methods can be useful, but they may not provide the same result for different soil and weather conditions. In recent years, machine learning and Internet of Things technologies have been increasingly used to support agricultural decision-making. This literature review studies existing research related to crop prediction, fertilizer recommendation, machine learning, IoT-based monitoring and intelligent agricultural decision-support systems. Different machine learning algorithms such as Random Forest, Decision Tree, K-Nearest Neighbour, Support Vector Machine, XGBoost, CatBoost, Logistic Regression and ensemble methods have been applied in previous studies. IoT systems have also been developed using sensors and microcontrollers to collect soil and environmental information in real time. The reviewed studies show that machine learning can provide useful predictions when sufficient and relevant agricultural data are available, while IoT can help overcome the limitation of relying only on static datasets. However, several limitations are still present in existing systems. Some studies focus only on crop prediction, while others concentrate only on fertilizer recommendation. Many systems are evaluated using small or region-specific datasets, and real-time sensor integration is not always included. Explainability, fertilizer quantity optimization, climate variation and practical deployment also require further attention. Based on these observations, this review identifies the need for an integrated system that combines crop prediction, fertilizer recommendation, real-time IoT data and comparative machine learning evaluation. The proposed research direction is therefore focused on developing a practical smart farming system that can use soil and environmental parameters to predict a suitable crop and recommend an appropriate fertilizer.