Does Jax’s linear regression model outperform sklearn’s predict_proba in a binary classification problem?

I have completed the first Kaggle playground competition for 2024 and found it to be a probability problem, which is something that I have seen in the past few Kaggle playground competitions in 2023. The link to the competition can be found here:- https://www.kaggle.com/competitions/playground-series-s4e1

In the past few Kaggle competitions where the task was to predict on the probability that a prediction will be a 1 I have used Jax’s linear regression model to achieve the probability. In this competition, however, I decided to use predict_proba in the sklearn dataset to compare it with Jax’s linear regression.

As can be seen from the screenshot below, I created four Jupyter Notebooks that I submitted from Kaggle. The notebook that performed the best was the one with the random forest model. The notebook that performed the least was the Jax notebook where I used the linear regression model:-

Since random forest performed the best amongst my submissions, I decided to carry out a review of the code.