Frequently asked questions regarding the use of machine learning models in the G2M platform

48 articles

What types of models can I build with the G2M platform?

How do I create a new model?

What variables should I select?

Which algorithm should I pick?

Which algorithm should I pick for my A/B testing analysis?

Which algorithm should I pick for my marketing mix model?

What is the estimated sample size?

When should I downsample my dataset?

What dataset training size should I use?

What is model drift?

How do I delete an existing model?

How do I share a model with another user?

What is the predict stage of a model?

How can I make predictions using an existing model?

How can I export predictions I just generated?

How do I interpret error metrics for my marketing mix model?

What is outlier removal?

How can I export my training results to a presentation?

How can I convert my G2M model to a Jupyter notebook?

How can I export my prediction results to a spreadsheet?

How can I duplicate a model?

How is prediction performance monitored?

What is the Infer stage of a model

Which algorithm should I pick for my propensity model?

What is SMOTE preprocessing?

How do I interpret driver rankings for my propensity or regression model?

How do I interpret error metrics for my propensity model?

What is a confusion matrix?

What is an ROC curve?

How do I interpret the propensity bin comparison chart?

How is prediction performance monitored for my propensity model?

Which algorithm should I pick for my regression model?

How do I interpret error metrics for my regression model?

How do I interpret coefficients for my regression model?

How do I interpret lag estimates for my regression model?

How do I interpret saturation estimates for my regression model?

How is prediction performance monitored for my regression model?