Generate Demo Data Set for Rides

[jamesward]

A ride consists of:

• A driver
• A rider
• Location of Driver
• Location of Rider
• Time to pickup
• Time of event

In order to do #1 we need a demo data set we can feed into PredictionIO. The data set can't be totally random otherwise our predictions might appear random. So maybe there is public taxi data we can use.

[anniexcheng]

This NYC Cab data set looks promising: http://www.nyc.gov/html/tlc/html/about/trip_record_data.shtml

[jamesward]

Cool. And I realized we don't really need the driver data for predicting demand, just the "request" or pickup data.

[adamwhat]

I'm making a sample of 2015 NYC Cab data set as the original data set is way too large for development. We can use the full data set to train the model once the whole pipeline is complete.

[adamwhat]

Sample Data Download

The sample is a 0.1% of the random sample with 1388068 entries from 2015 yellow cab dataset.

Here is the HDS5 version of the same data. You can use deepdish to load the data as a Pandas DataFrame.

[jamesward]

Fixed by #4

[adamwhat]

https://s3-us-west-2.amazonaws.com/4740/yellow_tripdata_2015_further_sample.csv.zip

sample of 10000 lines of taxi data 2015