quantitative-technologies / urbankit

An out-of-box analysis tool for urban sensory data

Urban Kit

An out-of-box analysis tool for urban sensory data.

Features

• Data Support

  • CSV File, including id,longitude,latitude columns and any attributes column.

• Statistic

  • Max/Min/Avg/Medium/Q1/Q3/Variance Dynamic Calculation.
  • Correlation Analysis: DTW/Pearson

• Learning Model

  • Timeseries Prediction: LSTM. LgihtGBM
  • Anomaly Detection: Islation Forest
  • Sptial Interpolation: SI-AGAN

Tech Stack

• Front end
  • Dev: Vue.js
  • Visualization: E-Charts
  • UI: Element-UI
• Back end
  • Dev: Django
  • Leanring Model: PyTorch, Scikit-Learn
• Databse
  • MongoDB

Design

• Front End

  

• Back End

  

Demo

• Core Operation

  

• Correlation Analysis

  

• Anomaly Detection

  

• Spatial Interpolation

  

• Timeseries Prediction

  

Reference

1. Yujia Gao, Liang Liu, Chi Zhang, Xiao Wang, Huadong Ma. "SI-AGAN: Spatial Interpolation with Attentional Generative Adversarial Networks for Environment Monitoring", In ECAI, 2020.
2. Ruiyuan Li, Huajun He, Rubin Wang, Yuchuan Huang, Junwen Liu, Sijie Ruan, Tianfu He, Jie Bao, and Yu Zheng. "JUST: JD Urban Spatio-Temporal Data Engine", In ICDE. IEEE, 2020.
3. Y. Zheng, L. Capra, O. Wolfson, and H. Yang. 2014b. "Urban computing: Concepts, methodologies, and applications". ACM Transactions on Intelligent Systems and Technology 5, 3 (2014), 38--55.
4. Liu, Fei Tony, Kai Ming Ting, and Zhi-Hua Zhou. "Isolation forest", 2008 Eighth IEEE International Conference on Data Mining. IEEE, 2008.