0
$\begingroup$

What is the difference between training data and features in a machine learning model?

Is feature just building blocks of training dataset?

$\endgroup$
1
  • $\begingroup$ Training data is the set ob observations and variables, usually in tabular form. Variables are also called features. $\endgroup$ Oct 10, 2019 at 8:19

1 Answer 1

1
$\begingroup$

Features are what your machine learning model will use as input, and they will be built or derived from the training data. At the end, given an instance of the data, you should be able to extract the features and pass it to the model for inference.

Lets say your training data is a set of messages (text) and you want to build a classifier that says wether the messages are important or not. Then, your Machine Learning models could take as input some TF-IDF representation for example, those would be the features. You could also add other feature to your model such as the Sentiment polarity as predicted from another machine learning model, or a binary variable stating wether the message contains some set of words.

So you can think of it as X: training data is mapped to X_features via some mapping. You can also think about the sklearn feature extraction interface which has the fit and transform methods and represents this mapping that takes you from the data to the features.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.