Recommender system

The client manages an educational platform that helps children to learn maths. It has several units and hundres of excersises, this in turn generates a lots of data that they seek to use to train a recommender system that can propose excersises to students based on their past performance.

Use the database with more of 10 thousand records to train and test a recommender system using neural networks.

Use Pytorch (a Python library) to create custom neuronal network in Python and train it with the given data.

Use R to calculate the best starting point for the neuronal network, use causal inference models.

Test the neuronal network with the rest of the data and report resuls on accuracy and precision using the cuadratic error metric.