Data Science II (course at WHZ)

The second semester of the data science lecture series starts with visualization techniques. Then supervised machine learning for generating predictions from data is introduced. Linear regression and artificial neural networks are discussed in depth.

Data Visualization

Week 1 (Matplotlib Basics)

Lectures

• Matplotlib

  • Matplotlib Basics

Self-study

• Matplotlib Basics (exercises; Circular Colorbar is bonus)

Practice session

• Chemnitz Trees: download, cleaning, short names (project)

Week 2 (Advanced Matplotlib)

Lectures

• Matplotlib, continued

  • 3D Plots

  • Animations

  • Seaborn

  • Maps

Self-study

• Advanced Matplotlib (exercises)

Practice session

• Chemnitz Trees: information extraction, presentation (project)

Week 3 (Ploty and Folium)

Lectures

• Plotly

• Folium

Self-study

• revisit Get Data and Set Up the Environment (project)

Practice session

• Public Transport

  • Find Connections (project)

Supervised Learning

Week 4 (Introduction)

Lectures

• Machine Learning Overview

• General Considerations

  • Problem and Workflow

  • Introductory Example (k-NN)

Self-study

Practice session

• Interactive Map (project, Color-Coded Distances are bonus)

Week 5 (Quality Measures and Scaling)

Lectures

• General Considerations, continued

  • Quality Measures (proof of AUC interpretation is bonus)

  • Scaling

Self-study

• revisit Climate Change (project)

Practice session

• Weather Animation (project)

Week 6 (Feature Reduction)

Lectures

• General Considerations, continued

  • Feature Reduction

Self-study

Practice session

• Forged Banknotes

  • Detecting Forgery with k-NN (project)

  • Quality Measures (project, AUC section is bonus)

Week 7 (Hyperparameter Optimization)

Lectures

• General Considerations, continued

  • Hyperparameter Tuning

Self-study

• MNIST Character Recognition

  • revisit/complete Load QMNIST (project)

• revisit/complete Image Processing with NumPy (exercises)

Practice session

• MNIST Character Recognition

  • QMNIST Feature Reduction (project)

Week 8 (Linear Regression)

Lectures

• Linear Regression

  • Approach

  • Regularization

  • Worked Example: House Prices I

Self-study

• Linear Regression, continued

  • Worked Example: House Prices II

Practice session

• Forged Banknotes

  • Hyperparameter Optimization (project)

• House Prices

  • House Prices GUI (bonus project)

Week 9 (More on Regression)

Lectures

• Linear Regression, continued

  • Outliers

• Logistic Regression

Self-study

• finish previous projects

Practice session

• finish previous projects

Week 10 (ANN Basics)

Lectures

• Artificial Neural Networks

  • ANN Basics

  • Training ANNs (mathematical derivation of ANN gradients and ANN implementation from scratch are bonus)

Self-study

Practice session

• House Prices

  • House Prices ANN

Week 11 (ANNs with Keras)

Lectures

• Cloud Computing

• Artificial Neural Networks, continued

  • ANNs with Keras

Self-study

Practice session

• PCA and ANN for QMNIST

• Long-Running Tasks

Week 12 (CNNs, part I)

Lectures

• Artificial Neural Networks, continued

  • Convolutional Neural Networks

  • CNNs with Keras

Self-study

Practice session

• CNN for QMNIST

Week 13 (CNNs, part II)

Lectures

• Artificial Neural Networks, continued

  • What did the CNN learn?

  • Improving CNN performance

Self-study

Practice session

• CNN Analysis for QMNIST

• Hyperparameter Optimization for Cats and Dogs (bonus)