These are some old kaggle notebooks of mine. The main objective was to first transform BTC data into images. Then using these images, build an image classifier.
I was wondering if I can obtain meaningful results by turning a time-series data into images first, and then classifying these images and trying to predict bitcoin trend for next day.
So every image contains weekly data of BTC price. Classifier tries to predict the next days value.
I couldn't achieve a meaningful accuracy so far.