AI-Powered Public Transport Route Optimization System

Every morning, thousands of passengers wait for a bus without knowing the exact arrival time. Traffic jams, accidents, breakdowns — a fixed schedule becomes fiction. We built an AI system that predicts passenger flow with 15-minute accuracy and dynamically adjusts headways. Result: wait times reduced by 20%, operational costs by 15%.

Stack: LightGBM, genetic algorithms, MILP optimization. All models run in real time on streaming data. Company metrics: 8+ years in ML production, 15+ transportation projects, 5+ years on the market. This article explains how we build predictive models, set up dynamic scheduling, and manage electric bus fleets.

What Are the Limitations of Traditional Methods?

Fixed schedules ignore real demand fluctuations. During peak hours, buses are overcrowded; at night, they run half empty. Dispatchers react after the fact, once a disruption has already occurred. AI analyzes data from AFC, GPS, cameras, and apps, builds forecasts, and reshuffles the schedule every 1-2 hours. This reduces wait times by 20% and increases load factor by 15%.

How Does AI Analyze Passenger Flows?

We collect data from multiple sources:

• Fare gates (AFC): exact entry/exit times, ticket type
• GPS trackers: real-time location, deviation from schedule
• In-vehicle cameras: YOLOv8 + tracking counts passengers
• Mobile app: geolocation with user consent

For prediction, we use LightGBM with lag features from 1, 7, and 14 days, moving averages, plus weather and holiday data:

import pandas as pd
import numpy as np
from lightgbm import LGBMRegressor

class PassengerFlowPredictor:
    """Predicts passenger flow at a stop for 15-minute intervals"""

    def build_features(self, df):
        df = df.copy()
        df['hour'] = df['timestamp'].dt.hour
        df['minute_bin'] = df['timestamp'].dt.minute // 15
        df['dayofweek'] = df['timestamp'].dt.dayofweek
        df['is_weekend'] = df['dayofweek'].isin([5, 6]).astype(int)
        df['month'] = df['timestamp'].dt.month

        # Lags: same intervals in previous periods
        for lag_days in [1, 7, 14]:
            df[f'lag_{lag_days}d'] = df['passengers'].shift(lag_days * 96)  # 96 intervals/day

        # Moving average
        df['ma_7d'] = df['passengers'].rolling(7 * 96).mean()

        return df

    def train_and_predict(self, historical_df, forecast_horizon=96):
        df = self.build_features(historical_df)
        feature_cols = ['hour', 'minute_bin', 'dayofweek', 'is_weekend', 'month',
                        'lag_1d', 'lag_7d', 'lag_14d', 'ma_7d', 'is_holiday',
                        'weather_temp', 'weather_rain']

        train = df.dropna(subset=feature_cols + ['passengers'])
        model = LGBMRegressor(n_estimators=300, learning_rate=0.05, num_leaves=64)
        model.fit(train[feature_cols], train['passengers'])

        # Forecast for next 24 hours
        future = df.tail(forecast_horizon)[feature_cols]
        return model.predict(future).clip(min=0)

The model trains on historical data and every 15 minutes outputs a forecast for the next 24 hours. Accuracy: MAE 3-5 passengers per stop.

Advantages of Dynamic Scheduling

Fixed schedules cannot adapt to demand fluctuations: peak-hour buses are overcrowded, night-time buses nearly empty. Dynamic scheduling adjusts headways every 1-2 hours based on forecasts. Optimal headway is calculated using the formula:

Dynamic scheduling is 1.3 times more efficient than fixed schedules, achieving a 1.5x improvement in load factor. When a vehicle breaks down, AI redistributes headways among remaining buses — passengers are unaware of the disruption.

Comparison of Headway Optimization Methods

Route Network Optimization

We use a genetic algorithm (GA) to find the optimal route configuration. Criteria:

• Coverage: 90% of residents within a 500 m walking distance of a stop
• Average number of transfers ≤ 2
• Minimize duplication of parallel routes

GA explores thousands of configurations in an hour — impossible with manual planning. For low-density zones, we deploy Demand Responsive Transport (DRT): the passenger requests a ride via app, the algorithm merges similar requests and builds a minibus route in real time (VRP solver). Additionally, we run transport simulation modeling on a digital twin of the city to evaluate network load.

How Is Fleet Management and Electric Bus Charging Optimized?

We optimize not only routes but also the number of vehicles in service. A MILP model minimizes empty trips from the depot while considering technical condition. For electric buses (LiAZ 6274, Yutong E12), we forecast energy consumption per route considering terrain and load. We build a charging schedule with electric bus charging optimization: up to 70% overnight at cheap rates, the rest at terminals. Guarantees sufficient charge before departure.

Implementation Steps

1. Audit of the current network and data collection (AFC, GPS, GTFS). 2-3 weeks.
2. Development of ML models (passenger flow forecast, headway optimization). 6-8 weeks.
3. Integration with city systems (traffic management, dispatch). 4-6 weeks.
4. Testing on historical data and pilot launch on 1-2 routes. 4 weeks.
5. Full network deployment, dispatcher training, monitoring. 4-6 weeks.

Total: basic platform — 4-5 months, with DRT and electric buses — up to 7 months. Typical project cost for a mid-size city: $200,000–$500,000, with annual savings of $2M–$5M. Cost is calculated individually.

What Our Work Includes

• Business analytics and route network audit
• ML models: forecasting, optimization, DRT
• Integration with GTFS, traffic management, mobile apps
• On-prem or cloud deployment
• Documentation, training, 6 months support
• Deliverables include: documentation, system access, training, and ongoing support

Our company has 8+ years of experience in ML production, completed 15+ transportation projects, and has been on the market for 5+ years. Our approach combines machine learning for transit with genetic algorithm route planning and electric bus charging optimization. Get a consultation: we will analyze your data, choose the stack, and propose a turnkey solution. Contact us for a project assessment.

Industry AI Solutions: Healthcare, Finance, Retail, Manufacturing

We encounter the same pain points: a general text model doesn’t distinguish medical nomenclature, and a standard object detector confuses “weld seam scratch” with “casing scratch.” Each time these are different defects with different consequences. To avoid this, we build industry-specific solutions on top of general methods, but with deep domain knowledge — from regulatory requirements to data specifics. Over 5 years, we have completed 80+ projects in fintech, healthcare, retail, and manufacturing, and none were without adaptation to a specific business case.

Healthcare: Regulatory Maze and Data Governance

Medical AI differs not in technical algorithms but in a compliance-first approach. Depending on the country of application, the model may be a Class II or III medical device requiring clinical trials (FDA, CE MDR, GOST R). We ensure compliance with these standards at the architecture stage — fixing them post-factum is 10× more expensive.

Medical imaging. Detection on X‑rays, CT, MRI is a mature area. Models on ResNet, EfficientNet, SegFormer achieve AUC 0.94–0.97 on standard tasks (pneumonia on CXR, polyps on colonoscopy). Key issue is generalization: a model trained on data from one scanner manufacturer degrades on another due to differences in preprocessing and artifacts. Solution: domain adaptation via MONAI (Medical Open Network for AI) from NVIDIA, which includes DICOM loading, 3D augmentation, and confidence calibration. TotalSegmentator — for automatic segmentation of 117 structures on CT, production‑ready, Apache 2.0 license.

Clinical NLP. Extracting structured information from clinical records: diagnoses (ICD‑10/11), prescriptions, dates, indicators. medspaCy, scispaCy, MedCAT — specialized NLP libraries with ontologies (SNOMED‑CT, UMLS). Fine‑tuning BioBERT or ClinicalBERT on our data yields F1 0.85–0.92 on NER tasks versus F1 0.65–0.72 for general BERT. We verified this on a project with a regional oncology center — cancer stage extraction accuracy increased by 23%.

Clinical decision support. LLM assistants for clinical decision support are a regulatory gray area. We use an RAG system on top of clinical guidelines (UpToDate, local protocols) with explicit citation for each statement. The model does not diagnose but helps find relevant protocols. Stack: LlamaIndex + pgvector + pubmedbert-base-embeddings + Llama Guard for safety. Data in DICOM/HL7 FHIR, on‑premise deployment mandatory.

Finance: How to Ensure Interpretability of a Scoring Model under Basel IV?

The financial sector is one of the most mature in applying ML, but regulation is maximal. Every model affecting credit decisions falls under Basel IV, EU AI Act, GDPR Article 22. We deliver AI solutions for fintech that satisfy these requirements — in a project for a top‑10 bank we deployed a scoring model where each record required SHAP explanations.

Credit scoring. Gradient boosting (LightGBM, XGBoost) dominates. Neural networks yield +0.5–2% AUC but lose interpretability. Standard: LightGBM + SHAP to explain each decision. Fairness checking is mandatory: Fairlearn or aif360 for auditing disparate impact on protected attributes (age, gender). The default class is 1–5% — with an imbalance of 1:30, a model with 97% accuracy may have recall 0.2. Solution: focal loss, class_weight='balanced', SMOTE + careful validation. In one fintech scoring project, the model reduced credit losses by $2.1 million annually.

Algorithmic trading and risk management. LSTM and Transformer for price forecasting are popular but unstable in production due to non‑stationarity of financial series. A more robust approach: ML for signal generation (classification: up/down over horizon N) with traditional portfolio optimization on top. Backtesting via Zipline‑Reloaded, vectorbt, QuantLib. Proper backtesting is critical — look‑ahead bias kills results. We guarantee a clean experiment: all data at signal time is available in real time.

AML (Anti‑Money Laundering). Graph Neural Networks for analyzing transaction networks is an actively developing area. PyG, DGL for GNN. Task: detect suspicious patterns in transaction graphs (layering, structuring). Recall is more critical than precision — better 10 false alarms than miss one money laundering. In a project for a large payment service, we increased recall by 18% without increasing false positive rate.

Retail and e‑commerce: Recommendation Systems and Demand Forecasting

Recommendation systems. Current architectural standard: two‑tower model for retrieval + ranking with cross‑features. TensorFlow Recommenders or Merlin from NVIDIA for GPU‑accelerated feature processing. For small catalogs (<100k items), LightFM is sufficient. A common mistake is training on implicit feedback without accounting for position bias. Solution: IPW (Inverse Propensity Weighting) or randomized logging on a portion of traffic. Development time for a basic recommendation system is 4–8 weeks, including A/B test.

Demand forecasting and inventory optimization. Hierarchical forecasting: SKU → category → store → region. HierarchicalForecast from Nixtla automatically reconciles forecasts across levels. TFT or N‑HiTS for base forecast, gradient boosting for adjustment on exogenous factors (promotions, weather, events). One retail project led to a 15% reduction in stock‑outs due to precise promotion calibration.

Visual search and size compatibility. CLIP embeddings for image search — deploy in 2–3 weeks: clip‑ViT‑B‑32 or clip‑ViT‑L‑14, Faiss or Qdrant index, REST API. For size recommendation — specific models on return data and reviews with fit indication.

Manufacturing: Quality Inspection and Predictive Maintenance

Quality control and defect detection. CV models for product inspection are one of the most mature industry tasks. YOLOv10 for defect detection, SegFormer for segmentation. Specifics: class imbalance (defects are rare), high recall requirement (missing a defect is worse than false alarm). Typical dataset: 500–2000 defect images + 500–1000 normal. Few‑shot learning via DINO or SAM 2 works with 50–100 annotated examples. We gained experience on an electronics production line — recall 0.95 at FPR 0.03. A predictive maintenance deployment saved a manufacturing client $500,000 per year in unplanned downtime.

Predictive maintenance. Vibration sensors, current sensors, thermocouples → feature extraction → anomaly or mode classification. Models: LSTM‑AE for unsupervised, LightGBM for supervised (if failure history is available). Integration with SCADA/OPC‑UA via opcua-asyncio or MQTT. Key metric: False Negative Rate — a missed pre‑failure is more costly than a false alarm. Threshold tuned to business cost of each error type. Timeline: 3 to 6 months to production.

Digital twin and simulation. Surrogate models — ML models replacing expensive physical simulation. If a CFD simulation takes 6 hours and a surrogate (trained on 10,000 simulations) takes 0.01 seconds, that's 2,000,000× speedup for optimization. SALib for sensitivity analysis, botorch for Bayesian optimization on top of surrogate.

General Principles of Industry AI

Regardless of industry, there are patterns that work everywhere. Data matters more than architecture. In healthcare, 1000 quality labeled images are better than 100,000 poor ones. In manufacturing, 200 real defect examples are more valuable than 10,000 synthetic ones. Compliance‑first design — regulatory requirements are easier to embed into architecture from the start than to add later. Logging, explainability, versioning from day one. Domain expert on the team — an ML engineer without domain knowledge does slowly and error‑prone what an ML engineer plus a doctor/financier/technologist does quickly and correctly.

We guarantee certification to customer requirements (ISO 13485, SOC 2, GDPR) and provide full model documentation (model card, datasheet, compliance report). Our experience: 10,000+ engineering hours and 80+ projects.

Work Process for an Industry AI Solution

1. Domain immersion (2–3 days) — interviews with experts, studying regulatory requirements, auditing available data.
2. MVP design (1–2 weeks) — stack and architecture selection, feasibility assessment.
3. Development and validation (from 4 weeks to 6 months depending on industry) — model training, testing, compliance.
4. Integration and deployment (1–4 weeks) — on‑premise / cloud / edge, documentation, staff training.
5. Support and monitoring — model drift, retraining, SLA.

Estimated timelines:

Cost is calculated individually for each project. Get a consultation — we will evaluate your dataset, regulatory map, and business goals.

Why Choose Our Industry AI Solutions?

• 80+ completed projects in fintech, healthcare, retail, and manufacturing.
• 5 years on the market — proven experience with compliance and deployment.
• Quality guarantee: we ensure target metrics (AUC, recall, latency p99) and provide full documentation.
• Licensed technologies: PyTorch, MONAI, LightGBM, Qdrant — we use open‑source with commercially safe licenses.
• Flexibility: we work as a contractor or as an extension of your team.

Contact us for a free data audit and consultation. Request a proposal with a detailed work plan. We will discuss your task and prepare a commercial proposal.