feature-eng | Implements classes for feature engineering | Machine Learning library

I am using the feature-engine library, and am finding that when I create an sklearn Pipeline that uses the SklearnTransformerWrapper to wrap a OneHotEncoder, I get the following error when trying to run cross-validation:

I have a dataframe with column grade which contains categorical values. My problem result in the fact, that the type of the values are float and not object.

I am trying to add pop_april_2010 values associated with each County to the map. However my code only returns the base map when I don't include the line "color='pop_april_2010:Q'". Including the line results in an empty image without the map.

Dataframe - df
The data is from Kaggle > https://www.kaggle.com/camnugent/california-housing-feature-engineering?select=cal_populations_county.csv

Code

For getting the list of installed libraries, I run the following command in Jupyter Notebook:

In the below-given code, I was implementing autofeat library for feature engineering. But when I ran the code in google colab, It gave an error, as stated below.

I want to install Feature-engine package on Kaggle to use YeoJohnsonTransformer.

I have some questions about feature-engine (reference):
1- to install the package, this document recommends:
pip install feature-engine

The feature-engine is with hyphen in the above code, however when calling it, it comes with underscore:
from feature_engine import variable_transformers as vt
Is that correct?

2- how can I fix the following error while installing the package?

ERROR: Could not find a version that satisfies the requirement feature-engine (from versions: none)

Thank you,

p.s. I could not create a new tag for feature-engine because" creating the new tag 'feature-engine' requires at least 1500 reputation".

I am working with feature_engine to fill missing values

I've implemented DoodleJump and want an AI to learn it via Q-Learning. The ANN and the whole Q-Learning process is already implemented as well.

But it appears to be not working so far. I'm pretty sure that the reason for that is located in the reward system or in the input feature-engineering (perhaps both).

So my question: What would be a good way to reward the agent here and which inputs should I give the ANN?

My approach was:

Reward system:

• reward -5 when the agent is too low, so that he loses
• reward 1 when hitting a platform
• otherwise the reward is always 0

Ihe inputs:

• the player looks to 8 directions (up, up-right, right, down-right, down, down-left, left, up-left)
• if there's a platform in a direction, the nearest one is chosen and the distance is calculated
• assume the distance is d. Then the input for this direction is: 1.0/d
• this way, the input is greater if the platform is nearer. And it is always smaller than 1
• if there's no platfrom to a certain direction, the corresponding input is 0
• the first parameter is the upmoving velocity of the agent (can be negative too) (also normalized, so it's always between -1 and 1)

So the whole input array: