when importing the alphatRNN.py file, i get the following error.

No module named 'keras.legacy'

Dear professors,

I believe there is a small issue in the Google_Colab_Setup.ipynb script.

The code defines a variable path, and later calls a variable dir_path, which is not defined and raises an error. I assume they are the same so it might just be a need to rename them.

Thanks,

Hi!
I'd like to use an extras example nr.5 for Chapter 3 but original link https://www.dropbox.com/sh/2dd1aida38nanwl/AACaCt1fXX2xr6I5T4B9U3_1a?dl=0 in notebook doesn't contain mtm_irs.npz datafile. Can you please advice where or how to find the file ? Thanks!

When running: conda env create python=3.6 -f C:\prosjekt\ML_Finance_Codes\environment.yml
or conda env create -f C:\prosjekt\ML_Finance_Codes\environment.yml

with anaconda 3, python 3.9.7 -64 bits, I get error installing xarray:

Requirement already satisfied, skipping upgrade: cached-property; python_version < "3.8" in c:\users\endre\anaconda3\envs\mlfenv\lib\site-packages (from h5py>=2.7.0->pymc3==3.8->-r C:\prosjekt\ML_Finance_Codes\condaenv.s7kg5d46.requirements.txt (line 2)) (1.5.2)
Requirement already satisfied, skipping upgrade: matplotlib>=3.0 in c:\users\endre\anaconda3\envs\mlfenv\lib\site-packages (from arviz>=0.4.1->pymc3==3.8->-r C:\prosjekt\ML_Finance_Codes\condaenv.s7kg5d46.requirements.txt (line 2)) (3.1.3)
Requirement already satisfied, skipping upgrade: xarray>=0.11 in c:\users\endre\anaconda3\envs\mlfenv\lib\site-packages (from arviz>=0.4.1->pymc3==3.8->-r C:\prosjekt\ML_Finance_Codes\condaenv.s7kg5d46.requirements.txt (line 2)) (0.18.2)

Pip subprocess error:
ERROR: Package 'xarray' requires a different Python: 3.6.15 not in '>=3.7'

failed

CondaEnvException: Pip failed

I'd like to start by saying that this isn't an issue with any of the code, rather an issue with the notation in the book. I didn't know where best to raise the issue directly so I thought this would be the most appropriate place.

I recently purchased this book and have been thoroughly enjoying it so far, I am currently on Chapter 6: Sequence Modelling. On page 194, section 2.2 Autoregressive Processes where it discusses the backshift operator and I came across notation for which I thought was confusing as it is written differently to my studies.

Just above equation 6.3, it describes the concept of the backshift operator as https://render.githubusercontent.com/render/math?math=y_{t-j}=L^{j}[y_{j}]
But surely this is wrong? As surely applying a backshift operation j and timestamp j, would just yield y0? And not y_t?

Wouldn't the correct notation be
https://render.githubusercontent.com/render/math?math=y_{t-j}=L^{j}[y_{t}]
Because yielding y_{t - j} would require the backshift operator of order j to be applied at y_t?

I apologise if I am wrong for this, but it just seems like something is off here.

Dear Mr. Matthew Dixon, and dear Mr. Paul Thalesians,

I really do enjoy the notebooks and want to make a suggestion if the writers plan a second part.
As there is several things about RL and IRL - maybe to consider a chapter about the Player of Games from Deepmind. An algorithm that excels in task with perfect and imperfect information. In my opinion this could be something for finance.

Now about my question. How is it possible to make a forecast for several steps with pm.Data and pm.set_data - so the shared variable - to make the Implementation of the stochastic volatility model with leverage in PyMC3? I am trying to arrange it myself but keep on hitting a wall.

Thank you very much in advance and best regards

Matthias

How to plug in the data into the Bayesian Neural Network, for example with yahoo finance data?
I have been trying around a lot and simply cannot get it together.
Please show a working solution.

This is my own shaky implementation up to now ..... but only in the sense of proof of concept.

https://github.com/waudinio27/Bayesian-Neural-Network-with-PyMC3

Thank you