Hands-On Deep Learning for IoT

Deep feature synthesis

Then, we generate the features using Deep Feature Synthesis (DFS). For this, we need to establish an entity set structure for our data. We can create an engines entity by normalizing the engine_no column in the raw data:

def make_entityset(data):    es = ft.EntitySet('Dataset')    es.entity_from_dataframe(dataframe=data,                             entity_id='recordings',                             index='index',                             time_index='time')    es.normalize_entity(base_entity_id='recordings',                        new_entity_id='engines',                        index='engine_no')    es.normalize_entity(base_entity_id='recordings',                        new_entity_id='cycles',                        index='time_in_cycles')    return eses = make_entityset(data)

The preceding code block will generate the following statistics of the entity set:

Entityset: Dataset   Entities: recordings [Rows: ...