The Station2vec framework consists of two components, for details please see the code:

• station corpus extraction
• station embedding.
  

The data utilized is the 30 days' transaction records of smart card in April 2015, which is provided by Shanghai Public Transportation Card Company. For the file of each days' records (more than 800 MB) exceeds the limits, so can not upload. All the data used is more than 23 GB.

Here are four records in the data:
卡号 日期 时间 站点 行业 金额
602141128 2015-04-01 07:51:08 1号线莘庄 地铁 0.0
602141128 2015-04-01 09:07:57 11号线昌吉东路 地铁 6.0
202608138 2015-04-01 10:27:06 无 轮渡 1.0
2603250687 2015-04-01 04:51:00 无 出租 21.0

• python (2.7.13)
• pandas (0.20.3)
• numpy (1.13.3)
• scipy (0.19.1)
• gensim (3.1.0)

The code is for the International Conference on Data Science(ICDS 2018) accepted paper: "Station2vec: Regional function discovery based on metro passenger flow data embedding"