jsleslie / ohare_taxi_demand Goto Github PK

Using project boards and tasks, make it clear who is responsible for what function. In particular, create an issue for each task.

Add documentation for the steps taken to extract hourly taxi data from the BigQuery taxi database

Include documentation on the steps taken as well as source

This issue is a reminder that the ride data we have doesn't account for the emergence of ride share companies in Chicago and their impact on taxi demand.

https://www.chicagotribune.com/business/ct-met-ride-hailing-numbers-20180201-story.html

How this is rectified is still TBD.

Baseline models

Feature engineering

H2O / SKLearn comparison

ORD_OTP dateset is missing flight data for April for all years 2013-2019 as well as for random months in each year. Appears to be an issue with the downloader since the FAA website has flights recorded in April.

• 1. Merge passengers2018raw.csv Aircraft type with L_AIRCRAFT_TYPE.csv found here for seat counts and airline and origin

• 2. Create a table/dictionary with Manufacturers, Models and Descriptions for step 3

• 3. Merge the Tail-Numbers from the plane_details.csv with the Aircraft type from the passengers2018raw.csv. Create a column which goes through each row in plane_details and adds a value corresponding to the appropriate aircraft description in the dictionary from step 2.

• 4. Merge with OTP to create master flight arrivals data

• ** Note that Tail-Numbers are not a complete set. For the ~20% of missing values, the average number of seats for the corresponding airline will be used to impute the number of seats.

Brainstorming issue for potential solution:

Include documentation for the steps taken to extract it from the US Bureau of Transportation