DataPredict™ [3 Years / Release 2.42] - Revenue & Game Optimization Using Machine Learning, Deep Learning And Reinforcement Learning (100+ Models)

Resources Community Resources
, , , ,
1176

Is this just a simple Markov chain for placement patterns, or are you actually running something more complex on the backend?

1177

More complex. If you read the tutorial, you would have noticed the usage of factorization machine.

Factorization machine is more commonly used for recommending content, but I have adapted it for placement recommendation. Best in cases where you have data with many zeros.

1179

TensorL-2D Library Update

Hello guys! I have made several improvements, changes and fixes to the tensor library used by DataPredict. This update includes:

  • Optimized matrix inversion operation. By default, this operation will use LU decomposition method for inversing the matrix when dimensionSize is greater or equal to six. When dimensionSize is lower than six, it would default to determinant method for inversing the matrix. You can override this behaviour by passing one of these strings as the second parameter: “LUDecomposition” or “Determinant”.

  • Fixed an issue related to the determinant method for inversing the matrix.

  • Optimized the overall codebase by changing on how the matrices are created.

  • Renamed a large number of variables for consistency within TensorL-2D’s API.

You can get the TensorL-2D’s library from here!

That’s all for today!

1183

Because of the update mention above, I have also updated the anti-cheat demo.

I also added toggles to simulate cheating. :3

Just get the demo from the main post.

See you all later!

1185

TensorL-2D Library Update

Hello guys! I have made several changes to the tensor library used by DataPredict. This update includes:

  • Renaming “Determinant” to “Adjugate” for choosing the matrix inversion method.

  • “LUDecomposition” method is now used by default for matrix inversion regardless of the matrix’s dimension size.

Apparently, I made a mistake when doing a time complexity analysis between the “Adjugate” method and “LUDecomposition” method for matrix inversion. So, I did a proper benchmarking instead. It was found that “LUDecomposition” method is more faster than the “Determinant” method just by looking at the calculation speed for getting the resulting matrix.

I recommend everyone to update the TensorL-2D library.

I apologize for the issues caused by my mistakes.

You can get the TensorL-2D’s library from here!

Thank you for your attention!

1 Like
1198

Hello guys!

I just added “Low-Risk Systems” section in the “High-Value Project Tutorials” documentation, which can be found here. I also have added the explanation for the worse case scenarios if these models gives you wrong or inaccurate predictions. With this, you will have a peace of mind knowing which models that will not impact your games in a very negative way!

That’s all for today and see you later!