How to prepare input data for machine learning project

Hello all.

I’m taking Machine Learning specialisation course, and so far all is good. However i cannot seem to understand how one would approach a Pattern detection project for Stock/Forex market in a give time window ( e.g 2021-12-31 - 2022-12-31) .

In the course, the instructor used handwritten digits as an example, and the data came pre-populated. (it looks like he took pixel by pixel RGBA value from the image and stored it into .npy file)

Now let’s look at this Stock market Pattern example…

  1. I have data which is the price of stock at each given time
  2. Data is a list of 2D vectors: [ [Time, Price], [Time, Price], ]

So question is how would i prepare and feed my input into a NN and detect as many Head and Should patterns/shapes (shown below) as possible in a Given time window?

if it’s covered in the course, you can tell me and i’ll move on with the course

Time-series models are not discussed in this course, they’re too complex for an introduction course.

Recurrent Neural Networks can handle this. They’re covered in Course 5 of the Deep Learning Specialization.

Thanks for your reply @TMosh

Let me finish the introductory course and get to the juicy part… :innocent: