Contents
The HSE created an algorithm to predict which of the homeless will be able to find a job and return home, and who will not. The charitable organization Nochlezhka helped the developers with this.
What’s going on
- Researchers from the Faculty of Computer Science (PCS) at the Higher School of Economics have developed a machine learning model that predicts with an 80% probability whether a homeless person will be able to undergo rehabilitation and leave the street.
- The model was trained on data provided by the charitable organization Nochlezhka. She uses the system of the Multifunctional Social Worker’s Office (MKS), where information about clients and services for them is entered. In total, there are 12 clients in the database. For training, information was used on 891 clients and 3 contracts – services for the homeless (temporary registration at the Nochlezhka address, restoration or receipt of a passport, and so on).
- First, the developers taught the model to predict the success of the contract. To do this, signs were identified by which the contact can be considered successful or unsuccessful.
- Each client in the dataset was described by 93 features in order to more accurately predict the probability of successful completion of the contract.
- The scientists plan to further develop the project: use ensemble methods and different architectures of neural networks, experiment with data obtained using computer simulations, and study data from other regions.
What does it mean
There is no exact number of homeless people in our country. Rosstat estimates their number at 64, the Nochlezhka charitable organization at 1,5–4 million. According to Nochlezhka, the number of homeless people is constantly growing.
In 2020, HSE sociologists studied the Moscow homeless and compiled a general portrait. This is a man from a city in the central part of our country, middle-aged, unmarried, with a secondary vocational education and work experience of two years.
Among the reasons why people ended up on the street, HSE and Nochlezhka single out moving to another city in search of work, family problems, fraudulent real estate transactions, job loss and the inability to rent housing.