Transformers in Spark's MLlib primarily implement the .transform() method. This method is essential for applying the transformation logic defined within the transformer to a DataFrame, resulting in a new DataFrame with the modified data. This functionality enables the chaining of multiple transformers in a pipeline for various data manipulations.

• .create(): Transformers are usually instantiated via class constructors or factory methods, not through a generic .create() method.
• .fit(): Although some transformers may involve a fitting process to learn parameters from data, .fit() is not universally applicable to all transformers.
• .append() and .combine(): These methods are more related to DataFrame operations rather than being specific to transformers.