Making Predictions
Explore the process of making sales predictions using a trained TensorFlow regression model. Understand how to prepare unlabeled test data, use input pipelines with TFRecords, and apply the Estimator predict function to generate accurate predictions for retail sales. This lesson helps you apply industry-relevant techniques to estimate outcomes from new data.
We'll cover the following...
We'll cover the following...
Chapter Goals:
- Use the regression model to make predictions on an unlabeled test dataset
A. The final model
After continuous training and evaluation of the sales prediction model, we found that the initial configuration of the model (2 hidden layers with 200 and 100 nodes, respectively) performs reasonably better than smaller models and has essentially the same performance as larger models. Therefore, we stuck with the initial configuration for the final model
The final model was ...