Machine learning workflow

As explained in Machine Learning Platform Engineering

The end-to-end sequence of activities through which a machine learning solution progresses from problem formulation and experimentation through deployment, operation, and retirement.

A machine learning workflow is the path from defining a useful problem to operating—and eventually retiring—a model. It is a loop, not a one-way checklist: changing requirements, data, or model performance can send a team back to earlier decisions. section 1.4.1 section 2.1

Use the phases to organize work, while allowing steps to be revisited, combined, or skipped when appropriate:

• Define the problem with stakeholders, then collect and prepare data for training and evaluation. section 2.1 section 1.5
• Version data, automate training in a pipeline, and track models and their supporting code so results remain reproducible. section 1.5 section 2.1
• Deploy the model for use, then monitor data and model behavior. Revisit earlier phases when performance or conditions change; retire models when appropriate. section 6.0 section 1.4.1

Production work needs more than a successful notebook experiment. Kubeflow Pipelines can connect stages as components that pass data downstream; feature stores can supply prepared features for training and serving. Deployment and monitoring then make the trained model an operating service whose behavior can be checked over time. section 1.3.3 section 6.0

When the loop turns back

If a model performs poorly during training, return to data preparation rather than treating deployment as the inevitable next step. A deployed used-car price predictor may likewise need new versions as customer needs and market conditions change. section 1.4.1 section 2.1

Figure 2.2 ML as a loop

The loop diagram reinforces that a model changes through repeated problem, data, and model decisions rather than a single pass. section 2.1

How Machine Learning Platform Engineering covers it

How deep it goes

Explains phases, iteration, and automation through production scenarios; leaves detailed implementation of individual tools to their respective treatments.

Examples

Uses a changing used-car price predictor, a training setback that prompts data rework, and an income-classification deployment.

What it assumes

Assumes software-engineering and basic ML-framework familiarity, but not expertise in algorithm internals such as backpropagation.

What you can do after

Practice/perform an iterative workflow that aligns a problem, automates repeatable stages, and responds to production changes.

How it connects

Connects data preparation, reproducibility, ML pipelines, model deployment, monitoring, data drift, and model lineage.

liveBookmanning.com

Read Machine Learning Platform Engineering on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Distributed Machine Learning PatternsThe stages and activities involved in developing, deploying, operating, maintaining, and retiring machine learning models and systems.
• 

  liveBookmanning.com
  Deep Learning with Python, Third EditionThe end-to-end sequence of steps for defining, building, evaluating, deploying, and maintaining a machine learning system.
• 

  liveBookmanning.com
  Graph Algorithms for Data ScienceA machine learning workflow is a sequence of processing and modeling steps for training or applying machine learning models.
• 

  liveBookmanning.com
  Deep Learning with R, Third EditionA machine learning workflow is an organized sequence of activities for defining a problem, preparing data, developing a model, and deploying and maintaining the resulting system.
• 

  liveBookmanning.com
  The Art of AI Product DevelopmentA machine learning lifecycle is an iterative process for preparing data, developing models, evaluating results, and applying machine-learning outputs.
• 

  liveBookmanning.com
  LLMs in ProductionA machine learning life cycle is the sequence of activities involved in creating, deploying, operating, and retiring machine-learning models.

Compare how 7 books explain machine learning workflow →

Related concepts

• pipelineequivalent to · A pipeline is an ordered or coordinated sequence of connected processing stages or tasks that passes data, work, or results toward an outcome.
• traininghas part · Model training is the process of fitting a machine-learning model to data by adjusting its parameters or structure to perform a task.
• monitoringhas part · Monitoring is the ongoing observation and measurement of a system’s behavior, state, or operational signals to assess its condition and detect changes or problems.
• preprocessinghas part · Preprocessing is the preparation or transformation of data, content, or other inputs into a form suitable for a subsequent computational operation.
• Model Servinghas part · A model-serving platform is software and infrastructure that hosts trained machine-learning models, handles inference requests, and returns model outputs.
• generationuses · Generation is the computational production of outputs from supplied inputs or context, often by applying a model to produce predictions or new content.

This explainer was generated by Manning's knowledge graph from the text of Machine Learning Platform Engineering. The section links point to the parts of the book it draws on.