Using SHAPELETS to categorize sensor time series data. The Learning Time-series Shapelets algorithm is used to transform the sequence dataset, then Gradient Boosting Classifier will be used to predict the label.
Learning Time-series Shapelets is an algorithm to learn shapelets. Its formulation is developed from logistic regression.
00:00 Introduction
01:05 Brief description of shapelets
02:24 First shapelet algorithms
08:25 Learning Time-series Shapelets: The modelling
11:38 Load the data and libraries
16:31 Convert data frames to 3D matrices
19:41 Split the data
20:39 Visualization of the sensors
21:28 Split data in more detail
28:58 Determine the total shapelets candidate
34:28 Implementation of Learning Time-series Shapelets
40:32 Shapelets plot
43:43 Training a Gradient Boosting Classifier model
46:26 Predict the test data
Source code: I'll upload it soon.
[Update] The source code is here: https://github.com/stephanielees/shapeletsLTS-humansensor
Thank you so much for watching!
#timeseries #classification #deeplearning #machinelearning #regression #logisticregression #timeseriesanalysis #datascience #python #kaggle