Time Series Classification in Python

Master time series classification in Python! This course covers machine learning and deep learning techniques for classifying time series, all applied in guided hands-on projects in 100% Python.

By the end of this course, you will:

• master time series classification

• perform feature engineering and model optimization for classification

• learn and implement state-of-the-art machine learning and deep learning models

• get hands-on experience with real-life datasets in the fields of healthcare, IoT, sensor data, spectroscopy and more

This is the most complete course on time series classification! We cover all types of models like:

• Distance-based

• Dictionary-based

• Ensemble models

• Feature-based

• Interval-based

• Kernel-based

• Shapelet models

• Meta classifiers

We first explore the theory and inner workings of each model before applying them in a hands-on project using Python.

Plus, get an additional section covering deep learning models, giving you a blueprint to apply any deep learning architecture for time series classification. All functions are flexible such that you can handle series with any number of features, samples and time steps.

Detailed outline:

• Introduction to time series classification
  

  • Application of time series classification

  • Baseline classifiers

• Distance-based method
  

  • Euclidean distance

  • K-Nearest Neighbors classifier

  • Dynamic Time Warping (DTW) from scratch

  • ShapeDTW

• Dictionary-based models

  • BOSS

  • WEASEL

  • TDE

  • MUSE

  • Capstone project: Japanese vowels' speakers classification

• Ensemble methods

  • Bagging

  • Weighted classifier

  • Time series forest

• Feature-based methods

  • Summary classifier

  • Matrix profile

  • Catch22

  • TSFresh

  • Capstone project: Classify equipment failure in a processing plant

• Interval-based method

  • RISE

  • CIF

  • DrCIF

• Kernel-based methods

  • Support vector machine

  • Rocket

  • Arsenal

  • Capstone project: Classify appliances by their electricity usage

• Shapelet-based methods

  • Shapelet transform classifier

• Hybrid models

  • HIVE-COTE

  • Capstone project: Beverage classification through spectroscopy

• EXTRA: Deep learning for time series classification

In this module, we develop a blueprint such that you can apply any deep learning architectures for time series classification. By the end, you will have built flexible functions that can adapt to series with any number of samples, features and time steps.

• Deep learning blueprint with Keras

• Deep learning blueprint with PyTorch