Classification (machine learning)

Good morning.

I need to implement a classification algorithm: I have several time series and I need to recognize the trend. For example, if I have the trend in the attached image, I want it to be recognised as ''type A'':

I have never done this kind of work. I know it is a supervised algorithm but that's all. Could you suggest me the procedure to follow or the best tools? I have read several articles, like such: https://www.mathworks.com/help/stats/classification.html

Of course, I would study the problem by myself. Right now, I really need few hints to start.

Thank you in advance

[Chris Basic]

I assume you have one graph that you want to call "Type A" and another graph with "Type B". If so, then you could have two columns, one with date (X-axis) and another with your Y-values. Add another column with classifying a data set as "Type A" or "Type B", convert it into a table, and use the classification learner app to run a certain classification method oof your choice.

Answer 1

[Bhargavi Maganuru]

For deep learning with time series, you can use LSTM network.

You can refer the following links to see how to train LSTM network and predict the new data.

https://www.mathworks.com/help/deeplearning/ug/time-series-forecasting-using-deep-learning.html

https://www.mathworks.com/help/deeplearning/ug/sequence-to-sequence-classification-using-deep-learning.html

[Veronica Taurino]

Hi, thank you. However I don't need to predict the evolution of the time series. I need a way to describe the time series as it is when I look at it. For this reason I thought about classification of pictures.

I am going to take a look to your suggestions, thank you