ML project for the price forecast of cars

The goal of this project was to predict the price of the car based on its technical features, text description and image.
Here is what was done:
A default ('naive') model was built as a benchmark for the future models
Exploratory Data Analysis was conducted to handle and normalize the features
The first model was built based on tabular data and the CatBoost algorithm
Additional linear models and their ensebles were used to enhance the results of the tabular model.
A simple dense neural network model was built
A multi-input neural network was built using both tabular data and pre-processed text
The images were added to the neural network
The final ensemble of Catboost and neural network was used to improve results
External dataset was uploaded in order to improve the model results.