Publicação
Forecasting real estate prices in Portugal based on a data science approach
| Resumo: | The evolution of the European residential market is notorious over the last ten years. House prices in the E.U. rose by 30.9 per cent between 2010 and the first quarter of 2021. The prices of homes in Portugal has risen almost 50 per cent during 11 years. Considering this previous argument, I propose the following research question: How to predict real estate prices. In this context, my research aims to analyze the prices’ evolution and understand the main components impacting the price of real estate. First, using the time series analysis, I use ARIMA to analyze the prices of real estate and the number of buildings sold since the first quarter of 2009, almost one year after the great recession in Portugal, until the third quarter of 2020 which was during the COVID-19 pandemic. The model was fitted and the prediction line was accurized with an upward trend. The second approach consists of analyzing the impact of five independent variables on real estate prices. To understand the most relevant components, regression analysis has been performed. I used OLS to analyze the impact of independent variables (crime rate, selected waste rate, tax rate, purchasing power and tourism rate) on real estate prices. Crime rate and tourism are negatively correlated while purchasing power, selected waste rate and tax rate are positively correlated with real estate prices. Then, I compared the accuracy of the result with neural networks and other types of regression analysis. Results were not much better than with linear regression. It is also essential to consider that this approach has some limitations, especially regarding the analysis’s granularity. The data has been collected from INE and PORDATA, the databases of contemporary Portugal, to construct the models and forecast house prices in Portugal. |
|---|---|
| Autores principais: | Samadani, Sanam |
| Assunto: | real estate prices prediction ARIMA Regression ANN OLS |
| Ano: | 2021 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório da Universidade de Lisboa |
| Resumo: | The evolution of the European residential market is notorious over the last ten years. House prices in the E.U. rose by 30.9 per cent between 2010 and the first quarter of 2021. The prices of homes in Portugal has risen almost 50 per cent during 11 years. Considering this previous argument, I propose the following research question: How to predict real estate prices. In this context, my research aims to analyze the prices’ evolution and understand the main components impacting the price of real estate. First, using the time series analysis, I use ARIMA to analyze the prices of real estate and the number of buildings sold since the first quarter of 2009, almost one year after the great recession in Portugal, until the third quarter of 2020 which was during the COVID-19 pandemic. The model was fitted and the prediction line was accurized with an upward trend. The second approach consists of analyzing the impact of five independent variables on real estate prices. To understand the most relevant components, regression analysis has been performed. I used OLS to analyze the impact of independent variables (crime rate, selected waste rate, tax rate, purchasing power and tourism rate) on real estate prices. Crime rate and tourism are negatively correlated while purchasing power, selected waste rate and tax rate are positively correlated with real estate prices. Then, I compared the accuracy of the result with neural networks and other types of regression analysis. Results were not much better than with linear regression. It is also essential to consider that this approach has some limitations, especially regarding the analysis’s granularity. The data has been collected from INE and PORDATA, the databases of contemporary Portugal, to construct the models and forecast house prices in Portugal. |
|---|