Our project aimed to create an interactive story map to enhance understanding and predict rental prices in Munich. By integrating visual storytelling, geospatial data, machine learning modeling, and a dynamic rental price calculator, we wanted to provide users with a practical tool to assess fair rental prices based on their preferences. We combined listing and spatial data to highlight key drivers of Munich’s rental market, enabling users to compare these factors with rental prices through interactive maps. The data also supported the deployment of a rental price calculator, which predicts prices based on various property features selected by the user.
Mapping and predicting rental prices in Munich
“Discovering the factors that influence rental prices in Munich.”
What it is about
How we built it
Our main tools for storytelling and interactive map visualization were ArcGIS Pro, Python, and JavaScript. First, data was collected from open-access sources like Open Street Maps for geospatial information and Kaggle for a large rental listing dataset (ImmoScout). The listing data was enriched with spatial data by applying spatial encoding techniques. Subsequently, machine learning models such as Linear Regression, Random Forest, and Gradient Boosting were trained and tested in Python to analyze the relationship between different variables and rental prices, and lastly make predictions. After evaluation, Gradient Boosting performed best with an RMSE of 268.7 EUR and an R2 score of 0.82, explaining 82,55% of the rental price variance.
Finally, ArcGIS Pro was used to style and integrate shapefiles into a story map. The model was deployed by creating a calculator in JavaScript, which was embedded into the story map so users could select options such as living space, property features, and location to predict rental prices interactively.
Finally, ArcGIS Pro was used to style and integrate shapefiles into a story map. The model was deployed by creating a calculator in JavaScript, which was embedded into the story map so users could select options such as living space, property features, and location to predict rental prices interactively.
Challenges we ran into
A key challenge in our project was data availability and quality. Despite gathering 3,843 rental listings along with spatial datasets, including eateries, city center, and parks, many records were discarded due to quality issues. We also recognize the critical importance of up-to-date data to ensure that the calculator remains aligned with Munich's dynamic rental market.
What we're proud of
We are proud of successfully developing and deploying the calculator that predicts rental prices in Munich. For us, our project demonstrates how technology, data analytics, and maps can be used to address real-world challenges and provide valuable insights for users.
Moreover, we are very pleased with our successful integration of various variables to provide a clear understanding of the drivers affecting rental prices in Munich. Through our story map, users can, for the first time, interact with a rental price map of Munich and explore how factors like proximity to the city center, amenities, and nearby facilities impact these prices.
Moreover, we are very pleased with our successful integration of various variables to provide a clear understanding of the drivers affecting rental prices in Munich. Through our story map, users can, for the first time, interact with a rental price map of Munich and explore how factors like proximity to the city center, amenities, and nearby facilities impact these prices.
What we learned
This project required skills in map design, machine learning, and web development. It certainly challenged our existing knowledge and allowed us to enhance our abilities in Python, JavaScript, and ArcGIS Pro. Additionally, we gained valuable experience in applying and deploying machine learning models to solve real-world problems.
What's next
Since we recognize the importance of up-to-date data for making accurate predictions, we plan to enhance the calculator in the future with updated rental data and additional features, such as demographic overlays. We also aim to showcase housing trends over time and include other relevant information about rental prices, which we currently lack.
Our Story Map is publicly accessible, and we hope it proves valuable to future intakes, especially those who are new to Munich.
Our Story Map is publicly accessible, and we hope it proves valuable to future intakes, especially those who are new to Munich.
Students
Betty Castro and Amilcar Figueroa
14th intake
Supervisor
Juliane Cron, M.Sc.
Keywords
Munich, Accommodation, Rental Prices, Machine Learning, Story Map