Neighborhood methods are a class of algorithms in machine learning and data science that make predictions based on the proximity of data points in the feature space. These methods assume that similar data points exist in close proximity, allowing them to infer properties or classifications based on the 'neighborhood' of a given data point.