One of the most challenging, but also potentially important, aspects of optimizing a model is choosing the values for the hyperparameters. In theory, we want to choose the best combination and, although we are unlikely to ever truly find the global maximum, the techniques in this section can help to find better values for the hyperparameters. Better hyperparameters can often improve the accuracy of a model.
Sometimes, however, a model has poor accuracy due to lacking the variables required for good prediction or because there is not enough data to support training a complex enough model to accurately predict or classify the data. In these cases, either acquiring additional variables/features that can be used as predictors and/or additional cases may be required. This book cannot help you collect more data, but it can present ways to tune and optimize hyperparameters. We'll deal with this next.