If you know which algorithm you need, you can go straight to that model and save time. Otherwise, use AutoLearn, as it determines which algorithm (or ensemble of algorithms) works best for your data.
All supervised models support both structured and unstructured data as input. Each takes a ground truth, plus at least one unstructured (free-form text) or training feature (a field that collects structured user data, that is, data that is numerical or that is selected from options). After the training fields are selected, the models pre-process the data you selected.
- For numerical training fields, a 0 is used to replace empty fields.
- For textual fields, it removes stopwords and lemmatizes and stems the text.
The main output is a label field that represents the Neural Sentiment Score. A keywords output field is added only if you select an unstructured field.
None of the Supervised models has an N-Gram as an output field. If you need N-Grams, please use the Unsupervised NLU model or the basic N-Gram Generator analysis.