Feature engineering

As explained in Deep Learning with R, Third Edition

Feature engineering is the manual design or transformation of input features into representations suitable for machine learning.

Feature engineering means using knowledge of the data and the learning algorithm to transform inputs before they reach a model. These transformations are hardcoded, not learned. The aim is to express the problem in a form that makes useful patterns easier to find. section 5.4.2

Reading a clock

Imagine predicting the time from a photograph of a clock. Raw pixels make this a demanding image-learning problem. Instead, a short R script could locate the tips of the hands and supply their (x, y) coordinates. A simple model could then associate those coordinates with times. Convert the coordinates to angles around the clock’s center, and even that model may be unnecessary: rounding and a lookup can recover the approximate time. section 5.4.2

Figure 5.17 Feature engineering for reading the time on a clock

The clock diagram shows how changing the input representation can simplify the prediction task. section 5.4.2

When developing a model:

• Filter out uninformative inputs through feature selection, then use problem knowledge to create promising new features. section 6.2.3
• Consider the features alongside the model architecture. A suitable representation helps direct learning toward useful solutions; a model may fail to find a simple solution even when it can represent one. section 13.2.3
• Keep data quality in view. Informative, less noisy features make the input space better structured and support generalization. section 5.1.2

Deep learning learns many representations that once had to be designed by hand, but this does not make input quality irrelevant. Nor is manual feature engineering the same as feature extraction from a pretrained model: that technique reuses representations the model has already learned. section 1.7 section 8.3.1

How Deep Learning with R, Third Edition covers it

How deep it goes

Defines hardcoded input transformations, works through a clock example, and gives workflow guidance without a general transformation recipe.

Examples

Uses clock-hand coordinates and angles to show how progressively better inputs simplify time prediction.

What it assumes

Introduces the idea through a familiar clock without requiring an account of neural-network internals.

What you can do after

Practice/perform feature selection and design non-learned input transformations using knowledge of the problem.

How it connects

Connects feature selection and generalization with architecture priors, learned representations, and pretrained feature extraction.

liveBookmanning.com

Read Deep Learning with R, Third Edition on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Machine Learning for Tabular Data: XGBoost, Deep Learning, and AIFeature processing is the transformation or elaboration of input features to make them suitable or more informative for machine learning models.
• 

  liveBookmanning.com
  Deep Learning with Python, Third EditionThe manual design or transformation of input variables into representations better suited for a learning algorithm.
• 

  liveBookmanning.com
  Knowledge Graphs and LLMs in ActionManual feature extraction is the hand-designed selection or computation of features from raw data.
• 

  liveBookmanning.com
  Natural Language Processing in Action, Second EditionFeature engineering is the deliberate construction or transformation of input representations to make them useful for predictive modeling.
• 

  liveBookmanning.com
  Outlier Detection in PythonFeature engineering is the process of modifying or combining original data attributes to create inputs that are more useful for analysis or modeling.
• 

  liveBookmanning.com
  Machine Learning System DesignFeature engineering is the process of constructing, transforming, or selecting input features for a machine learning model.
• 

  liveBookmanning.com
  Effective Data AnalysisFeature engineering is the creation or transformation of variables from existing data to better represent processes relevant to a model.
• 

  liveBookmanning.com
  Financial AI in PracticeFeature engineering is the process of constructing informative input variables from raw data for statistical or machine-learning models.
• 

  liveBookmanning.com
  Delivering Machine Learning Projects: From design to deploymentFeature engineering is the transformation of raw data into informative input variables that represent useful structure for a machine learning model.
• 

  liveBookmanning.com
  Sutskever's ListFeature engineering is the process of designing input representations from raw data for use by a learning algorithm.
• 

  liveBookmanning.com
  The Art of AI Product DevelopmentFeature engineering is the process of creating, selecting, transforming, and refining input features to improve a machine-learning model's results.
• 

  liveBookmanning.com
  Distributed Machine Learning PatternsFeature engineering is the process of creating, transforming, or selecting input features to improve a machine learning model.
• 

  liveBookmanning.com
  Digital Twins in ActionFeature engineering is the process of constructing informative input variables from raw data for analysis or machine learning.
• 

  liveBookmanning.com
  Evolutionary Deep Learning: Genetic algorithms and neural networksFeature engineering is the construction, transformation, or selection of input variables that represent useful information for a machine-learning model.
• 

  liveBookmanning.com
  Demand Forecasting Best PracticesFeature engineering is the selection, creation, or modification of input features for a machine learning model.

Compare how 16 books explain feature engineering →

Related concepts

• foundational modeluses · A base or foundation model is a pretrained model that provides learned parameters or capabilities as a starting point for reuse or adaptation to downstream tasks.
• feature selectionhas part · Feature selection is the process of choosing a subset of input features for use in a modeling, analysis, or prediction task.
• data layerpart of · A data layer is an architectural part of a system that organizes access to, management of, or persistence of data for other components.
• pipelinepart of · A pipeline is an ordered or coordinated sequence of connected processing stages or tasks that passes data, work, or results toward an outcome.
• language modeladdressed by · A language model is a computational model that represents patterns in linguistic tokens to estimate sequence probabilities and predict or generate language from context.
• signalproduces · A signal is an observable value, event, or pattern that conveys information about a condition, state, or phenomenon.

This explainer was generated by Manning's knowledge graph from the text of Deep Learning with R, Third Edition. The section links point to the parts of the book it draws on.