MUTra‑CDMX: Multisource Urban Traffic for CDMX

MUTra-CDMX v4.0 is a multisource urban traffic data package containing a complete five-minute spatiotemporal record for 20 consecutive road segments located along a 14.72-km section of the Insurgentes Sur corridor in Mexico City. The dataset covers 120 uninterrupted days, from 1 November 2024 at 00:00:00 to 28 February 2025 at 23:55:00, with 288 timestamps per segment per day. The complete package is distributed as MUTra_CDMX_v4.0.rar. Its primary analytical file, data/MUTra_CDMX_completed.csv, contains 691,200 unique segment–timestamp observations and 12 variables. Each row represents the traffic state and associated meteorological conditions of one road segment at a specific five-minute timestamp. The primary file includes: two spatiotemporal identifiers: segment_id and datetime; four traffic variables: current_speed, free_flow_speed, current_travel_time, and free_flow_travel_time; three meteorological variables: temperature_c, precipitation_hourly_total_mm, and wind_speed_ms; two derived operational indicators: relative_speed_index and congestion_index_0_10; one reconstruction-status variable: is_imputed. The final time grid contains no missing segment–timestamp combinations. Of the 691,200 records, 682,264 correspond to originally retrieved observations and 8,936 correspond to reconstructed combinations added during temporal completion. Original traffic observations were preserved without modification, while reconstructed records are explicitly identified through is_imputed. Hourly meteorological observations were aligned with five-minute traffic records using the interval [h,h+1)[h,h+1)[h,h+1) in the America/Mexico_City time zone. Temperature and wind speed are repeated across the twelve five-minute timestamps associated with each hour. precipitation_hourly_total_mm represents the total precipitation reported for the complete hour and should not be summed across repeated five-minute rows. Static physical, geometric, infrastructure, and urban-context attributes are provided separately in metadata/segment_metadata.csv. These attributes include segment coordinates, nominal length, functional road class, endpoint elevations, endpoint gradient, sinuosity, consolidated signalized-location counts, and nearby educational and healthcare-related points of interest. The spatial ordering and adjacency of the road segments are documented in metadata/segment_topology.csv. These files can be joined to the primary dataset through segment_id. The package also contains: a data dictionary and source documentation; record-level reconstruction provenance; weather-alignment documentation; point-of-interest and signalized-location provenance; temporal-completeness and integrity audits; artificial-masking validation results for the reconstruction procedure; predictive-utility benchmark materials; a file manifest and supporting documentation. The dataset supports travel-time forecasting, congestion analysis, mobility characterization, imputation studies, temporal sequence modeling, and the evaluation of machine-learning methods for Intelligent Transportation Systems. Users should consult the included README.md, data dictionary, provenance files, validation outputs, and licensing documentation before reuse.