Supervised machine learning is based on the ability of an algorithm to train a model using examples. A supervised machine learning algorithm needs certain enabling conditions to be met in order to perform. These enabling conditions are as follows:

• Enough examples: Supervised machine learning algorithms need enough examples to train a model.
• Patterns in historical data: The examples used to train a model need to have patterns in it. The likelihood of the occurrence of our event of interest should be dependent on a combination of patterns, trends, and events. Without these, we are dealing with random data that cannot be used to train a model.
• Valid assumptions: When we train a supervised machine learning model using examples, we expect that the assumptions that apply to the examples will also be valid in the future. Let's look at an actual example. If we want to train a machine learning model for the government that can predict the likelihood of whether a visa will be granted to a student, the understanding is that the laws and policies will not change when the model is used for predictions. If new policies or laws are enforced after training the model, the model may need to be retrained to incorporate this new information.