AI Residential Housing Stock Management System

Developing AI System for Residential Housing Stock Management

Management of residential housing stock at regional or municipal level is a task of balancing resources among thousands of properties with different technical conditions. AI transforms planning of repairs and investments from intuitive level to analytical.

Assessment of Building Technical Condition

Housing stock registration + ML:

Assessment data: construction year, project series, wall material, area, number of floors, repair history, current technical inspection. ML regression on aggregate wear index:

import pandas as pd
import numpy as np
from sklearn.ensemble import GradientBoostingRegressor

def build_wear_model(housing_data, inspection_results):
    """
    Building physical wear forecast from registry data.
    inspection_results: technical inspection results (target variable)
    """
    features = [
        'age_years',           # building age
        'wall_material_code',  # 0=panel, 1=brick, 2=monolithic
        'project_series',      # series (Khrushchyovka, Brezhniovka, etc.)
        'floors',
        'area_sqm',
        'basement_area_sqm',
        'last_major_repair_years_ago',
        'roof_material_code',
        'heating_type_code',
        'facade_cladding',
    ]

    X = housing_data[features]
    y = inspection_results['wear_pct']  # % physical wear

    model = GradientBoostingRegressor(n_estimators=300, learning_rate=0.05)
    model.fit(X, y)
    return model

Computer vision for inspection:

Facade photographs → CV condition assessment:

• Cracks: YOLOv8 Instance Segmentation → crack area and width
• Plaster peeling, balcony collapse, leaks
• Condition classification: satisfactory / unsatisfactory / emergency

Drones with cameras for roof inspection — inaccessible areas without scaffolding.

Major Repair Planning

Short-term risk:

Model of emergency failure forecast (probability over 3 years):

• Features: current wear, degradation rate, incident history
• LightGBM binary classification → P(emergency_3y)
• Priority: high risk + large population → first in queue

Regional major repair program:

Optimization of 10–25 year plan (per RF Housing Code Art. 168):

import pulp

def optimize_repair_program(houses, risk_scores, annual_budget, n_years=10):
    """
    Major repair program optimization: maximize preserved fund value
    under annual budget constraints.
    """
    prob = pulp.LpProblem("repair_program", pulp.LpMaximize)

    # Variable: in which year to repair house i
    repair_year = {}
    for house_id in houses['id']:
        for year in range(n_years):
            repair_year[(house_id, year)] = pulp.LpVariable(
                f"repair_{house_id}_{year}", cat='Binary'
            )

    # Each house repaired exactly once (or not at all)
    for house_id in houses['id']:
        prob += pulp.lpSum(repair_year[(house_id, y)] for y in range(n_years)) <= 1

    # Annual budget
    for year in range(n_years):
        prob += pulp.lpSum(
            houses.loc[houses['id']==h, 'repair_cost'].values[0] * repair_year[(h, year)]
            for h in houses['id']
        ) <= annual_budget

    # Objective: maximize weighted sum (risk × residents × 1/year)
    prob += pulp.lpSum(
        risk_scores[h] *
        houses.loc[houses['id']==h, 'residents'].values[0] *
        (1 / (year + 1)) *  # earlier repair preferred
        repair_year[(h, year)]
        for h in houses['id'] for year in range(n_years)
    )

    prob.solve(pulp.PULP_CBC_CMD(msg=0))
    return {(h, y): repair_year[(h, y)].value() for h in houses['id'] for y in range(n_years)}

Monitoring and Response

Incident management:

Integration with dispatch service (GIS HCS, Domovyk, 1C:HCS):

• Classification of resident complaints by type (NLP on complaint text)
• Routing to appropriate service
• Prioritization: water leak on 5th floor → urgent; squeaky door → planned

Analytics for municipality:

• KPI of fund condition: average wear, % emergency buildings
• Resource requirement forecast (relocations from emergency housing) for 3–5 years
• Financial model: contribution receipts vs. program funding need

Development timeline: 3–5 months for fund condition assessment system, major repair planner, and analytics dashboard.