This thesis covers two subjects regarding the real estate market in Prague. In the first part, we look for factors that influence the differences between offer and realized prices of residential properties. From our dataset, we identify the area and the time spent on market as the variables with the largest impact on the price differences. Additionally, we find that price differences are spatially correlated and tend to influence each other. Finally, accessibility of the apart- ment to given POI's seems to have a small but significant effect as well. In the second part, we build a neural network to predict the transaction prices per meter squared. After thorough architecture adjustment and hyperparameter tuning, we propose a model which is able to improve the current best prediction on the dataset by more than 12 %.
| Identifer | oai:union.ndltd.org:nusl.cz/oai:invenio.nusl.cz:452625 |
| Date | January 2021 |
| Creators | Kalous, Václav |
| Contributors | Polák, Petr, Komárek, Luboš |
| Source Sets | Czech ETDs |
| Language | English |
| Detected Language | English |
| Type | info:eu-repo/semantics/masterThesis |
| Rights | info:eu-repo/semantics/restrictedAccess |
Page generated in 0.0028 seconds