Featuring-data

EDA Plots:
- New algorithm for density scatterplots that finds how many points are within a defined radius of each point. It is slow, so it will only be used when there are less than 10,000 points.
- Another new algorithm for density scatterplots, this one for when there are more than 10,000 points. This algorithm uses np.histogram2d with enough bins, so that each actual point is consistent with one of the bins. Then, for each bin, add up the points in adjacent bins.
- Updates to plot functions to give option for creating plot without saving it.

EDA:
- Added new EDA function for compiling summary statistics on each column.

Other:
- Updates to __init__ files in order to access individual functions.
- Created new 'utils' folder to keep utility functions that could be useful outside the package as well.

Updates to the EDA functionality:
- EDA: Updated (reduced) the number of estimators/trees when using RF for correlation. This change produces essentially the same results in testing on different datasets, but will run faster, especially for larger datasets.

Updates to the XGBoost recursive model training:
- Added 3 new class attributes: the LabelEncoder object, the list of feature columns for the best iteration, and the xgboost model object for the best model.
- Training: Due to sklearn deprecation of the 'squared=False' option for 'mean_squared_error', the code will now first try 'root_mean_squared_error' (which only appears in newer versions of sklearn). If the code cannot import this, then it will fall back to existing 'mean_squared_error'.

Training: Bug fix related to changing 'test' names to 'val'.

New features:
- Print to terminal / Jupyter notebook tables of correlation values while the EDA code is running.
- Display in Jupyter notebook plots of the six highest correlated features (to the target variable) for numeric and non-numeric features.

Some small modifications to the plots, to keep a consistent theme. Additional bug testing.

An early release, while checking minimum dependency requirements and additional bug testing.

For this release, the code was tested on older versions of python and the required libraries. The `project.toml` file now includes minimum version requirements of all dependencies.