Modeling¶

The sections described below provide information to help you easily navigate the ML modeling process.

Overview¶

Topic	Describes...
Fundamentals of modeling	Understand the types of ML modeling projects you can create in DataRobot. Learn the general process of modeling, analyzing, and selecting models for deployment.
DataRobot workflow overview	A quick overview of steps from data ingest to model deployment. Each step links to the full documentation for more complete detail.

Topic	Describes...
Build models	Elements of the basic modeling workflow.
Advanced options	Setting advanced modeling parameters prior to building.
Manage projects	Manage models and projects, and export data.

Topic	Describes...
Evaluate tabs	View key plots and statistics needed to judge and interpret a model’s effectiveness.
Understand tabs	Understand what drives a model’s predictions.
Describe tabs	View model building information and feature details.
Predict tabs	Make predictions in DataRobot using the UI or API.
Compliance tabs	Compile model development documentation that can be used for regulatory validation.
Comments tab	Add comments to assets in the AI Catalog.
Bias and Fairness tabs	Identify if a model is biased and why the model is learning bias from the training data.
Other model tabs	Compare models across a project.

Topic	Describes...
Date/time partitioning	Build models with time-relevant data (not time series).
Unsupervised learning	Work with unlabeled or partially labeled data to build anomaly detection or clustering models.
Composable ML	Build blueprints using built-in DataRobot tasks and custom Python or R code.
Visual AI	Use image-based datasets.
Location AI	Use geospatial datasets.

Topic	Describes...
What is time-based modeling?	The basic modeling process and a recommended reading path.
Time series workflow overview	The workflow for creating a time series project.
Time series insights	Visualizations available to help interpret your data and models.
Time series predictions	Making predictions with time series models.
Multiseries modeling	Modeling with datasets that contain multiple time series.
Segmented modeling	Grouping series into segments, creating multiple projects for each segment, and producing a single combined model for the data.
Nowcasting	Making predictions for the present and very near future (very short-range forecasting).
Enable external prediction comparison	Comparing model predictions built outside of DataRobot against DataRobot predictions.
Advanced time series modeling	Modifying partitions, setting advanced options, and understanding window settings.
Time series modeling data	Working with the time series modeling dataset: Creating the modeling dataset Using the data prep tool Restoring pruned features
Time series reference	How to customize time series projects as well as a variety of deep-dive reference material for DataRobot time series modeling.

Topic	Describes...
No-Code AI Apps	Build and configure AI-powered applications using a no-code interface.
Value Tracker	Measure success by defining expected, and tracking actual, model value in real time.

Topic	Describes...
Data and sharing	Dataset requirements, sharing assets, and permissions.
Modeling details	The Leaderboard and the processes that drive model building, including partitioning and feature derivation.
Eureqa advanced tuning	Tune Eureqa models by modifying building blocks, customizing the target expression, and modifying other model parameters.