The backbone of all Machine Learning algorithms is the data. Everything a machine learning algorithm learns is from the data. Therefore it is critical to provide the correct data to the algorithm which is representative of the problem statement. As seen already, deep learning in particular requires large amounts of data for training models. We can sometimes say that a certain amount of data is enough for a problem, however there is never enough! More is better. The complexity of the model that is able to be trained correctly is directly proportional to the amount of data on which it is trained. Limited data will put an upper limit on the choice of model architecture for the problem. When considering the amount of data available, it is also worth noting that a portion of this will also need to be used for validation and testing purposes.

The following section will now discuss the data partitioning and its importance on the progress of any machine learning task.