Weight a specific feature Is it possible to manually add weight to a specific feature in Machine Learning? I have 51 features: 1 of the 51 features should have the same weight as the other 50 features.
This is because one feature is an animal name (7 different animals types) (text), the 50 other features describe statistics (mean, sd, skewness etc.) of the spoken animal names (speech). The output is a 1 or 0. 1 if text and speech describe the same animal, 0 if text and speech do not describe the same animal.  
 A: Technically, it is possible to give more weight to a specific feature:


*

*For random forests, you could manipulate the probability of selecting the feature, or, barring access to the internals of the algorithm, simply duplicate the feature so that its multiplicity reflects the weight you want it to have. For regularized logistic regression, you could give lower regularization penalties for this features. Different methods will be appropriate for different algorithms.

*As a more general method, if you perform dimension reduction using PCA before using another algorithm, you could normalize the feature differently than the other features.

*As a very general method, you could use two soft classifiers: one using the first feature, and one using all other features, and then take the average.
However, you might want to check (using cross-validation) that these methods outperform classifiers simply operating on all features regularly. For one, your intuition about the weight of the feature might be incorrect. Also, some form of weighting features is what learning algorithms already do.
