Time series Classification - README

Problem Description

The task correctly classify samples in the multivariate time series format. In other words, since this is a classification problem, the objective is to correctly map the information contained in the features calculated over time to their labels.

Dataset Details

You can download the dataset here

• Time series shape: 2429x36x6

• File format: .npy

• Number of Classes: 12

• Classes:

  0. "Wish"
  1. "Another"
  2. "Comfortably"
  3. "Money"
  4. "Breathe"
  5. "Time"
  6. "Brain"
  7. "Echoes"
  8. "Wearing"
  9. "Sorrow"
  10. "Hey"
  11. "Shine"

How to Use the Code

1. Ensure you have the required Python libraries installed, especially TensorFlow, Keras, and Keras Tuner.

2. Open the provided Python code in a compatible environment (e.g., Jupyter Notebook, Google Colab).

3. Run the code step by step, following the order described in the code.

Report

For more details about the model you can read the report, where everyting is explained in detail.

Conclusion

This code is designed to address the time-seris classification task with the provided dataset for ANNDL course challenge here at polimi. My team have received 5.5 out of 5.5 points doing this challenge.