How AI and Big Data could change real estate valuation
2022
Machine Learning
Full-Stack
Next.js
Python
My bachelor thesis is dedicated to the research and evaluation of artificial intelligence for the automated estimation of real estate prices in Germany.
Problem statement
Real estate valuation is complex and based on a variety of factors that affect the price of real estate.
Artificial intelligence (AI) has the ability to recognize such complex relationships in an automated manner.
Fundamentals of real estate valuation
There are two important concepts in real estate valuation: price and value.
Price is the amount paid when selling or buying a property.
Value, on the other hand, refers to the true worth of a property based on various factors such as location, size, condition and amenities.
Implementation of the work
In this bachelor thesis, Big Data and AI were used to automate real estate valuation. First, large amounts of data from real estate sales in Germany were processed and analyzed.
Then, different models were trained using machine learning to estimate real estate prices.
Results
The best estimates were obtained using an eXtreme Gradient Boosting Regression Tree, which has a median deviation of 81,773 euros from the actual property price.
The median of the absolute deviations was 45,527 euros.
In the results, the performance of machine learning was found to be superior to multiple linear regression.
Feature Importance
In machine learning, "feature importances" are a measure of how important certain features are for the prediction of a model.
They can be used to understand and optimize the model by focusing on the most important features and removing or modifying less important features.
Typically, feature importances are expressed as values between 0 and 1, with 1 being the highest value.
A feature with a value of 1 means that it is particularly important for predicting the model, while a value of 0 means that it has no influence.
Summary
The results of this bachelor thesis show that AI is a suitable technology to automate real estate valuation.
Practical applications could be automated real estate valuation in the form of a support system for real estate agents and for initial appraisal by property owners.
Related Links
This article is a very short summary of my bachelor thesis and contains only the a fraction of the information contained in the thesis.
- Written bachelor's thesis
- Web application (The web application does not use real data, but only static example data)
- Grade
- 15 out of 15 points
- Lines of code
- 1,403
- Pages
- 69
- Graphics
- 26
About me
I'm a software developer from Germany. I've been working with software development for more than 12 years. I'm passionate about technology and I love to learn new things. I'm currently working as a senior software developer at a company called "Engel & Völkers Technology".