Mentat-IT offers consulting and hands-on services in data science & machine learning

Data science & machine learning problems can often be broken down into a distinct number of tasks using the CRISP-DM Methodology.

CRISP-DM Process Diagram

1. Business Understanding

Translate your business problem into a data problem

  • Understand project objectives and requirements from a business perspective;
  • Converting this knowledge into a data problem definition;
  • Preliminary plan designed to achieve the objectives.

2. Data Understanding

Provide insight into your data

  • Data collection
  • Identify data quality problems
  • Discover first insights into data

3. Data Preparation

Prepare your data for machine learning

  • Transformation of data into usable formats
  • Feature selection – which properties are relevant to the problem?
  • Cleaning of data
  • Recursive process with modeling phase

4. Modeling

Apply various modeling techniques, depending on the problem e.g. supervised learning, unsupervised learning, deep learning.

  • Selection of a model suitable to the problem
  • Parameter selection – tuning the algorithm
  • Recursive process with data preparation phase

5. Evaluation

Test and evaluate the model to see if the business objectives have been achieved.

  • Testing model performance on test data
  • Evaluate whether outcomes fits the business objectives
  • Recursive process with data preparation & modeling phase

6. Deployment

Bring insights and results from the modeling phase into a production stage.

  • Reporting and visualisation of results
  • Implement the model into production ready code