Data Preprocess Source and Custom Dataset

[cpzz50]

This question is about BTC high frequency trading dataset provided.

From the csv file. Some columns like [bid1_price] has number [23090.7], but other columns like [high] have number [-0.8507737401414914]. Seems like some columns are normalized but others are not. Please advise how these data are preprocessed so I can put my own data into training.

Also about preprocessing, README indicate that the data is from Kaggle, which I found the original data doesn't contain indicators in the csv file. Wondering if it's possible disclose how those technical indicators are preprocessed.

Thank you

[qinmoelei]

EarnHFT contains a more detailed description of the data preprocess, which members also develop from TradeMaster. For high-frequency trading for crypto, the two repos share the same setting, and EarnHFT is much easier to read.