Skip to content

🧠 Traffic Profile Analysis 📦

Status Badge Module Type Internal Usage

This module cleans, processes, and clusters time series traffic data for integration into simulation tools.

Author: Belén López Salamanca Date: [11-07-2025]
Version: 1.0.0

🔍 Data Collection

Traffic data was captured from https://traficosevilla.es/, which provides hourly traffic counts for various monitored routes in the city of Seville. The data spans a full 24-hour period and was exported in two segments: morning (00–12h) and afternoon (12–24h).

🧼 Data Cleaning and Preparation

The raw CSV files were read and cleaned using Python. Key steps included:

  • Replacing missing values ('-') with zeros (Later, those were deleted for clustering)
  • Merging the two halves of the day into a single unified 24-hour profile per route.
  • Joining metadata such as route names from a separate CSV file containing route identifiers and street names.

There are 90 main roads determined by the Ayuntamiento but only the data of 46 roads are saved!

🧠 Time Series Clustering

We applied K-Means clustering to the time series data of vehicle counts per hour per route.

  • The series were normalized to allow pattern-based grouping rather than scale-based.
  • A total of 3 clusters were chosen based on visual validation and interpretability. At first glance it was clear that the data was split in 3 groups.

📈 Clustered Time Series and Cluster Description

Cluster Description
0 Medium traffic routes with regular daily rhythm
1 High-traffic routes with sharp morning peaks
2 Low-volume

Below is the plot of all routes, colored by cluster membership. Each line represents a route's hourly traffic profile. Thicker lines show the average pattern for each cluster. Cluster Time Series

🗺️ Map of Routes

The following map visualizes the spatial distribution of the routes, overlaid with cluster color coding:

Route Map

📍 Cluster Membership by Route

Cluster 0

RUTAS Calle
1 PALMERA_ENTRADA
2 PALMERA_SALIDA
5 LUIS_MONTOTO_ENTRADA
6 LUIS_MONTOTO_SALIDA
7 KANSAS_CITY_ENTRADA
8 KANSAS_CITY_SALIDA
9 JIMENEZ_BECERRIL_ENTRADA
10 JIMENEZ_BECERRIL_SALIDA
11 CRISTO.EXPIRACIÓN_ENTRADA
12 CRISTO.EXPIRACIÓN_SALIDA
18 TAMARGUILLO_SUR.-_NORTE-SUR
21 VIRGEN_DE_LUJAN_-_SUR-NORTE
22 VIRGEN_DE_LUJAN_-_NORTE-SUR
24 R.HISTORICA-_NORTE-SUR
25 PASEO_COLON_-_NORTE-SUR
26 PASEO_COLON_-_SUR-NORTE
45 AVDA._PAZ_ENTRADA
46 PUENTE_DE_LAS_DELICIAS_ENTRADA
47 PUENTE_DE_LAS_DELICIAS_SALIDA
50 TAMARGUILLO_NORTE_-_SUR-NORTE
51 TAMARGUILLO_NORTE_-_NORTE-SUR
53 TORNEO_-_SUR-NORTE
59 MARTINEZ.BARRIOS_-_SUR-NORTE
89 MENENDEZ_PELAYO_-_SUR-NORTE
90 MENENDEZ_PELAYO_-_NORTE-SUR

Cluster 1

RUTAS Calle
72 R.NORTE(SECTOR_ESTE)_-_NORTE-SUR
73 R.NORTE(SECTOR_OESTE)_-_ESTE-OESTE
74 R.NORTE(SECTOR_OESTE)_-_OESTE-ESTE
81 CTRA._UTRERA_ENTRADA

Cluster 2

RUTAS Calle
13 REP._ARGENTINA_ENTRADA
14 REP._ARGENTINA_SALIDA
17 TAMARGUILLO_SUR.-_SUR-NORTE
19 SU_EMINENCIA_-_SUR-NORTE
20 SU_EMINENCIA_-_NORTE-SUR
30 BORBOLLA_-_NORTE-SUR
32 EDUARDO_DATO_SALIDA
33 DR._FEDRIANI_ENTRADA
34 DR._FEDRIANI_SALIDA
38 RONDA_TRIANA_-_NORTE-SUR
39 RONDA_TRIANA_-_SUR-NORTE
42 CTRA._CARMONA_SALIDA
43 CTRA._CARMONA_ENTRADA
78 CARLOS_III_-_SUR-NORTE
83 RONDA_PIO_XII_-_OESTE-ESTE
84 RONDA_PIO_XII_-_ESTE-OESTE
87 CARDENAL_ILUNDAIN_-_OESTE-ESTE
88 CARDENAL_ILUNDAIN_-_ESTE-OESTE

🔮 Next Steps

  • 🧹 Interpolation: Fill in any minor gaps or missing values more intelligently than simple zeroing.
  • 🔁 Data enrichment: Coordinate with David to access historical or higher-resolution data.
  • 🚗 Integration with SUMO: Translate cluster-based traffic profiles into SUMO simulations using flows or OD matrices.

📎 Notes & References

  • Raw traffic data: https://traficosevilla.es/
  • Preprocessing and clustering done using pandas, scikit-learn, and matplotlib.
  • Optional interactive visualization made with plotly.