1. Distributions of variables begin to mirror those in the observed data

We can compare the distribution of a variable, like body mass index (BMI), in the simulated data to that in the observed data. Drag the epoch slider below to compare the distributions over increasing training epochs.

2. Correlations between variables begin to match real-world patterns

We can compare the correlation between variables in the simulated data to that in the observed data over increasing training epochs. Drag the epoch slider below to observe how the colors in the correlation heat map in the simulated data more closely resemble those in the observed data.

3. The model’s goodness of fit steadily improves

Drag the slider below to observe how the model’s goodness of fit improves over increasing training epochs.

To learn more about how Analysis Group’s data science capabilities can support your work, please reach out to aiblog@analysisgroup.com.