Applied Machine Learning

In the module "Foundation in Machine Learning", the basis is created for examining data with supervised and unsupervised learning methods.

In practice, the question arises as to how these methods should ideally be used. Or more concretely: How are data and features prepared to successfully develop predictive models? Which metrics are to be selected for a specific practical use case? How can models be compared, their predictions plausibilized and explained?

The module "Applied Machine Learning" is therefore less about algorithms and their implementation from scratch, but about the correct use and practice-oriented combination of machine learning methods.

Students are able to define target variables to answer domain-specific questions, construct additional variables with predictive power using feature engineering and unsupervised learning approaches, and partition data to enable model verification and productive use of models.

Students are able to identify practice-relevant metrics for evaluating model performance and correctly apply standard methods for regression and classification tasks. Furthermore, they can use frameworks and methods to train, compare and optimize predictive models and finally they know strategies for the development of machine learning models.

Students will be able to quantify model outputs with descriptive statistics (e.g., the distribution of model scores), analyze them with unsupervised learning methods, and identify the influence of explanatory variables.

Finally, students will be able to estimate the effect of models for pilot testing and will have an understanding of how to monitor data and model quality for model application in production.

Exploratory Data Analysis, Data Wrangling, Foundations in Machine Learning