Training process without cross-validation

[ycseong07]

Currently, the training process in stove requires the v-fold ('v' in models functions) parameter to be set and it's minimum value is 2. However, since the workflow cannot be applied to data with a small number of rows, an additional workflow that does not perform CV must be implemented.

[ycseong07]

Cross-validation is an essential step in the workflow and is always implemented due to the following reasons:

• The 'stove' package has an 'AutoML' concept that automatically optimizes the hyperparameters of each model.
• Hyperparameter optimization aims to create a generalizable model that does not overfit the training data and performs well on new datasets.
• To achieve this goal, hyperparameter optimization needs to be combined with logic that prevents overfitting, and cross-validation is the process that achieves this goal.
• Furthermore, the amount of data required for ML model training depends on the complexity of the task and learning algorithm, usually requiring hundreds or more data.
• However, if too little data is used for training, the model may not perform well. To prevent this, we recommend using at least 1,000 data points, as mentioned in the README.md file.