how to produce X.npy and Y.npy files

[lyw2000china]

Hi, @AlfredCYL . I like your project because I am planning to do something similar. It would be helpful if you could share some code files about data preprocessing. For example, how to produce X.npy and Y.npy files. Thank you.

[AlfredCYL]

Consider you only have the open, close, high, and low prices as your inputs. For each of them, you have a dataframe for n days and p stocks. The forward returns have the same shape. Then, for Y.npy, you just need to transform it into a numpy array (simply use pd.DataFrame.values()). For X.npy, you should process each data point in the same way as Y, and then use the np.concatenate function to merge them into the final numpy array.

I hope it helps. Feel free to ask if you still have problems.

[leonyvon]

Consider you only have the open, close, high, and low prices as your inputs. For each of them, you have a dataframe for n days and p stocks. The forward returns have the same shape. Then, for Y.npy, you just need to transform it into a numpy array (simply use pd.DataFrame.values()). For X.npy, you should process each data point in the same way as Y, and then use the np.concatenate function to merge them into the final numpy array.假设您只有开盘价、收盘价、最高价和最低价作为输入。对于它们中的每一个，您都有一个 n 天和 p 个股票的数据框。远期回报具有相同的形状。然后，对于 Y.npy，您只需将其转换为 numpy 数组（只需使用 pd.DataFrame.values()）。对于X.npy，您应该以与Y相同的方式处理每个数据点，然后使用np.concatenate函数将它们合并到最终的numpy数组中。

I hope it helps. Feel free to ask if you still have problems.我希望它有帮助。如果您还有问题，请随时询问。

For the Y, I wonder the index are symbols or columns are symbols?

[AlfredCYL]

Consider you only have the open, close, high, and low prices as your inputs. For each of them, you have a dataframe for n days and p stocks. The forward returns have the same shape. Then, for Y.npy, you just need to transform it into a numpy array (simply use pd.DataFrame.values()). For X.npy, you should process each data point in the same way as Y, and then use the np.concatenate function to merge them into the final numpy array.假设您只有开盘价、收盘价、最高价和最低价作为输入。对于它们中的每一个，您都有一个 n 天和 p 个股票的数据框。远期回报具有相同的形状。然后，对于 Y.npy，您只需将其转换为 numpy 数组（只需使用 pd.DataFrame.values()）。对于X.npy，您应该以与Y相同的方式处理每个数据点，然后使用np.concatenate函数将它们合并到最终的numpy数组中。 I hope it helps. Feel free to ask if you still have problems.我希望它有帮助。如果您还有问题，请随时询问。

For the Y, I wonder the index are symbols or columns are symbols?

Sorry for the late response. Y is (n_dates, n_stocks); X is (n_dates, n_features, n_stocks).