| Title |
Predicting real estate prices with machine learning / |
| Translation of Title |
Mašininio mokymosi taikymas nekilnojamojo turto kainų prognozavimui. |
| Authors |
Agadagba, Blessing Oritsewenyimi |
| Full Text |
|
| Pages |
69 |
| Keywords [eng] |
Keywords: Real Estate Price Prediction, LightGBM, Ensemble Learning, XGBoost, Linear Regression, Feature Engineering, Machine Learning, Vilnius Municipality Raktiniai žodžiai: Nekilnojamojo turto kainų prognozavimas, LightGBM, Ansamblio mokymasis, XGBoost, Linijinė regresija, Savybių inžinerija, Mašininis mokymasis, Vilniaus savivaldybė |
| Abstract [eng] |
This master's thesis explores a range of machine learning algorithms, including linear regression, Ridge, Lasso, ElasticNet, Decision Tree, Random Forest, XGBoost, LightGBM, and Gradient Boosting, for predicting real estate prices using data from Vilnius Municipality. The study outlines the steps for data cleaning and preparation, modelling, and hyperparameter tuning. The research follows a systematic approach, beginning with a baseline model using linear regression on apartment data alone. Additional demographic variables and newly engineered features were incorporated to assess their impact on model performance. The models were tuned and evaluated using three distinct configurations of the dataset: (1) apartment data alone, (2) apartment data combined with demographic variables, and (3) apartment data augmented with both demographic variables and newly engineered features. The results reveal that ensemble and boosting models significantly outperformed linear models, with LightGBM achieving the highest predictive accuracy (R² = 0.8698, RMSE = 35,714, MAPE = 12.54%) when apartment data with both demographic and newly engineered features were utilised. |
| Dissertation Institution |
Vilniaus universitetas. |
| Type |
Master thesis |
| Language |
English |
| Publication date |
2025 |
