Test performance on Kaggle

Results on train set: 0.304 RLMSE, on test set: 0.75094 RLMSE and 451 place(for now)

Upgrade performance.

Reducing learning time by changing the algorithm for transferring pre-processing from CPU to GPU

Add Optuna for auto search params.

v1.4

Update README.md

Improved model predict.

Upgrade model hyperparameters with SearchCV.

First run XGBoost.

Data analysis, visualization, pre-processing and construction of the first model without changing standard parameters were performed.