Part V. GIGO. Deep Learning Trading Strategy from the beginning to the production.
This is the part V of the Deep Learning Trading Strategy from the beginning to the production series.
In this video we are going to have a look at my fails. We will see what is wrong with highly imbalanced dataset, why ROC curve is not a good metric if your data set is imbalanced, and how the model performs in case of garbage data.
Note: As we are interested in profitable trades, we can replace all -1 and 255 labels with 0. After that we get binary labels and we can try to solve a binary classification task.
00:20 Critique of TensorFlow
01:10 Highly imbalanced dataset
02:28 Imbalanced data + ROC
03:45 ROC Curve
03:57 AUC
05:04 Garbage In, Garbage Out (GIGO)
05:34 Conclusion
Files from the video: https://github.com/CloseToAlgoTrading...
File IVE_tickbidask.zip with raw data you can find here: https://github.com/CloseToAlgoTrading...
Смотрите видео Part V. GIGO. Deep Learning Trading Strategy from the beginning to the production. онлайн без регистрации, длительностью часов минут секунд в хорошем качестве. Это видео добавил пользователь CloseToAlgoTrading 21 Март 2020, не забудьте поделиться им ссылкой с друзьями и знакомыми, на нашем сайте его посмотрели 1,664 раз и оно понравилось 35 людям.