Introduction
Explore how to select and implement a multilayer perceptron model to predict weekly sales for retail stores. Understand data preparation, model training, evaluation, and making predictions on new data within an industry-level machine learning project.
We'll cover the following...
In this Lab you will be building the machine learning model for predicting sales. Then, after training and evaluation to create the final model, you’ll use it to make sales predictions on new data.
A. Which model do we use?
Now that we’ve analyzed and processed the project’s data, we need to choose the correct model for the task. While certain applications benefit from more advanced models (e.g. CNNs for image recognition, LSTMs for NLP), a majority of use cases can be handled by a simple multilayer perceptron.
Our case study deals with predicting weekly sales for various retail stores. The feature data used in the model’s input pipeline consists of regular numeric and categorical features, meaning that an MLP will work just fine in making predictions.