This project was designed to scrape the contents of the Slovenia most popular website for real-estates - nepremicnine.si, process the data and design an ML model to predict the prices based on the characteristics of apartments for the city of Ljubljana, because it is the largest in the country and has substantially more data to work on. There are several files in this project.

‍

‍

The developed model aimed to serve as a predictive tool for rental prices in Ljubljana, leveraging machine learning and advanced hyperparameter tuning. The goal was to streamline the process of predicting rental costs with precision, eliminating manual efforts in hyperparameter optimization. The model underwent an innovative approach to hyperparameter tuning, utilizing automated methods for informed searches. The optimal combination of hyperparameters, including 64 units, a dropout rate of 0.3, stochastic gradient descent as the optimizer, and a learning rate set at 0.001, was identified and implemented.

‍

Upon model execution with the optimized hyperparameters, the predictive capabilities were significantly enhanced. The refined model showcased improved accuracy in forecasting rental prices based on key apartment characteristics in Ljubljana. This automated approach not only expedited the hyperparameter tuning process but also yielded a more efficient and accurate rent predictor for the specific real estate market in Slovenia's capital city.

‍