Using Technology to Prevent Tenant Evictions
A new framework aims to improve outreach for tenants at risk of eviction.
Anindya Sarkar, Alex DiChristofano, Sanmay Das, Patrick J. Fowler, Nathan Jacobs, Yevgeniy Vorobeychik
― 6 min read
Table of Contents
- Understanding Tenant Evictions
- The Need for Effective Outreach
- Introducing Active Geospatial Search (AGS)
- How AGS Works
- The Challenge of Prediction
- Building the Framework
- Evaluating AGS
- Key Findings
- Related Research
- Active Search Techniques
- Visual Active Search
- The Hierarchical Approach
- Evaluating Performance
- Variations in Query Costs
- Combining Data Types
- Conclusion
- Original Source
In many cities, tenant Evictions pose a significant challenge to housing stability. The ongoing situation can be a lot like playing a game of Jenga-one wrong move, and the whole thing comes crumbling down. To help those at risk of eviction, there’s a pressing need for effective Outreach programs. This is where data-driven methods come into play.
Our aim is to find out if using smart technology can make these outreach programs better. We propose a new framework called Active Geospatial Search (AGS) that focuses on identifying rental units needing help while taking into account travel costs and limited Resources.
Understanding Tenant Evictions
Tenant evictions can throw people into a whirlwind of instability, especially those in marginalized communities like families with children or single moms. When a family is evicted, it can lead to a domino effect that worsens the housing crisis, impacting not only the tenant but also the overall rental market. The available data shows that evictions in the U.S. are rising, with millions of cases filed every year. This is particularly concerning because it affects those who can afford it the least.
During the COVID-19 pandemic, a temporary relief in eviction rates was seen due to moratoriums at various levels of government. However, as these measures have been lifted, we are back to facing the challenge of rising eviction rates.
The Need for Effective Outreach
One way to address the issue of evictions is to provide resources directly to tenants at risk. Information about legal representation, financial assistance, and other resources could be vital to help tenants stay in their homes. However, canvassing neighborhoods to reach these tenants can be labor-intensive and the resources available are often limited.
So, how do we make the most of what we have?
Introducing Active Geospatial Search (AGS)
This is where our Active Geospatial Search framework comes in. AGS is designed to help outreach workers efficiently seek out Households that may be at risk of eviction. Think of it as a treasure hunt, but instead of treasure, the goal is to find people who need help.
The AGS framework identifies a sequence of properties to investigate, predicting which units are at higher risk of eviction. It’s all about making the best use of time and resources while adapting to new information as it comes in.
How AGS Works
AGS uses something called a hierarchical reinforcement learning approach. This fancy phrase essentially means it learns from experience, adjusting its strategies based on what works and what doesn’t.
Imagine you’re on a scavenger hunt with a budget. You can only visit a certain number of houses, and each one costs you time and energy. AGS figures out which houses are most likely to have people needing help and directs you there, all while keeping track of your budget.
The Challenge of Prediction
One of the biggest hurdles in this process is that we don’t know upfront which households are at risk. We can use historical data to guess, but predictions can quickly become outdated.
This makes it important to balance two approaches: exploration (gathering new information) and exploitation (using information we already have to find at-risk households). AGS is designed to find that balance effectively.
Building the Framework
In AGS, we set up various locations in a geographical area, each representing a rental unit. The system uses a search policy to decide which houses to check out first based on factors like past eviction filings and other property details.
Each time a location is checked, a cost is incurred, which can vary depending on the distance traveled. AGS’s primary goal is to maximize the number of discoveries while staying within the budget.
Evaluating AGS
To see how well AGS works, we evaluated it using eviction data from a large urban area. Results showed AGS is considerably more efficient at identifying eviction cases compared to traditional methods.
Key Findings
- Efficiency: AGS outperformed baseline methods, making it a valuable tool for outreach.
- Adaptability: The search policy can react to new information about evictions as it comes in.
- Budget Management: AGS manages resources efficiently, ensuring maximum outreach with limited funds.
Related Research
AGC fits into a larger body of research focused on using technology to tackle social issues. Similar methods have been applied in different areas like disaster relief, food donation redistribution, and more. However, none have focused specifically on eviction prevention in this way.
Active Search Techniques
AGS builds on existing active search techniques, which are used to find specific data points in a large dataset. While traditional methods look for instances with known labels, AGS needs to explore without prior knowledge about the target properties.
Visual Active Search
A related model, called Visual Active Search (VAS), uses images to help identify target objects in a large image area. However, AGS doesn’t rely solely on visual data since it focuses on the geographical context of properties.
The Hierarchical Approach
To enhance AGS further, we introduce a hierarchical structure that divides the search area into smaller regions.
- Level 1 Policy: This policy decides which larger region to investigate based on a set of inputs.
- Level 2 Policy: Once a region is selected, this policy zeroes in on specific properties to check.
This setup allows for better management of complex searches in large areas.
Evaluating Performance
When we tested AGS on actual eviction data, it consistently outperformed traditional methods, especially in areas with limited resources.
Variations in Query Costs
We also looked at how different query costs influenced performance. In scenarios where finding targets was more difficult, AGS showed even more promising results, highlighting its adaptability.
Combining Data Types
One interesting aspect of our work was how visual and tabular data influenced performance. While visual data can provide insights, tabular data often carries more weight. However, when used together, they create a more powerful tool for identifying at-risk households.
Conclusion
The Active Geospatial Search framework represents a significant step toward improving outreach efforts for tenants at risk of eviction. By effectively blending reinforcement learning with geospatial data, AGS is like having a secret weapon in the fight against housing instability.
As we move forward, it’s essential to consider not only how these technologies can help but also the ethical implications of using data to serve vulnerable populations. With the right approach, AGS could greatly assist social service agencies in connecting tenants with the help they need, potentially sparing them from the chaos that evictions often bring.
Let’s hope the only thing that gets evicted is the outdated way of finding at-risk tenants!
Title: Active Geospatial Search for Efficient Tenant Eviction Outreach
Abstract: Tenant evictions threaten housing stability and are a major concern for many cities. An open question concerns whether data-driven methods enhance outreach programs that target at-risk tenants to mitigate their risk of eviction. We propose a novel active geospatial search (AGS) modeling framework for this problem. AGS integrates property-level information in a search policy that identifies a sequence of rental units to canvas to both determine their eviction risk and provide support if needed. We propose a hierarchical reinforcement learning approach to learn a search policy for AGS that scales to large urban areas containing thousands of parcels, balancing exploration and exploitation and accounting for travel costs and a budget constraint. Crucially, the search policy adapts online to newly discovered information about evictions. Evaluation using eviction data for a large urban area demonstrates that the proposed framework and algorithmic approach are considerably more effective at sequentially identifying eviction cases than baseline methods.
Authors: Anindya Sarkar, Alex DiChristofano, Sanmay Das, Patrick J. Fowler, Nathan Jacobs, Yevgeniy Vorobeychik
Last Update: Dec 19, 2024
Language: English
Source URL: https://arxiv.org/abs/2412.17854
Source PDF: https://arxiv.org/pdf/2412.17854
Licence: https://creativecommons.org/licenses/by-nc-sa/4.0/
Changes: This summary was created with assistance from AI and may have inaccuracies. For accurate information, please refer to the original source documents linked here.
Thank you to arxiv for use of its open access interoperability.