Tuning stock models for better performance
Some machine learning tasks are well suited to be solved by the stock models presented in prior chapters. For these tasks, it may not be necessary to spend much time iterating and refining the model, because it may perform well enough without additional effort. On the other hand, many real-world tasks are inherently more difficult. For these tasks, the underlying concepts to be learned tend to be extremely complex, requiring an understanding of many subtle relationships, or the problem may be affected by substantial amounts of random variability, which makes it difficult to find the signal within the noise.
Developing models that perform extremely well on these types of challenging problems is every bit an art as it is a science. Sometimes a bit of intuition is helpful when trying to identify areas where performance can be improved. In other cases, finding improvements will require a brute-force, trial-and-error approach. Of course, this...
