From Albatross to MATSim plans


Author: Konstanze Winter


1.Albatross Data  Quick Overview
	 only represents 1/5 of population
	 data set for the entire Netherlands (Albatross Model for the year 2004): 89.663.504 activities performed by 2.193.772 x 5 agents living in 1.367.411 x 5 households  Focus on Amsterdam: Postcode 1000  1109
	 in 118.900 household at least one agent performs at least one trip from/to/within Amsterdam. In total the data set, 787.534 x 5 activities are performed from/to/within Amsterdam by 183.563 x 5 agents


2.From Albatross to plans for MATSim_AmsterdamScenario  Step by Step

Step 1: Focus on Amsterdam
The data set is reduced to households where at least one household member is traveling from/to/within Amsterdam (postcode 1000-1109).

Step 2: Add person ID and population Data
Based on the activity numbers, individual agents per household are distinguished

Step 3: Add Geo-Codes
Based on the list of postcodes (retrieved in 2017 from http://www.sqlblog.nl/nederlandse-postcodetabel-sql-script-voor-import/), geocodes are added to the data.
1.	Delete agents (not households), which have at least one activity not linked to a postcode (concerned 1184 agents). Replace non-existing PPC with closest existing PPC (six cases, PPC 2081?2082 and PPC 6679?6678)
2.	Per activity: From the list of postcodes, choose a random address with matching 4-digit code. 
3.	Add geo-coordinates as stated in list of postcodes to the activity
4.	Per agent: Set geo-coordinates of all home activities equal to the one of the first home activity
5.	Per agent: Set geo-coordinates of all work activities equal to the one of the first work activity, if they have the same 4-digit postcode than the first work activity  
6.	Change geocodes from GPS geo-codes to (wgs84) to EPSG32631 - in Java code: 
		Coord coordWSG = new Coord(actTemp.longitude, actTemp.latitude); 
		CoordinateTransformation ct=TransformationFactory.getCoordinateTransformation  (TransformationFactory.WGS84, TransformationFactory.WGS84_UTM31N); 
		Coord coord = ct.transform(coordWSG);

Step 4: Calibrate Modal Shares
11 simulations of the same day to obtain the modal split observed for Amsterdam in 2015 (see: https://assets.amsterdam.nl/publish/pages/905215/atb_2019.pdf)

Step 5: "Phase 1"
introduce SAV , as mode choice option
76 simulations of the same day (agents' plans from step 4 as input) to let agents include SAV

Step 6: "Phase 2"
16 simulation of the same day (agents' plans from step 5 as input). Necessary step to avoid oscillation (see: https://github.com/matsim-org/matsim-code-examples/issues/148#issuecomment-501765476)

Step 7: "Phase 3"
2 simulations of the same day (agents' plans from step 6 as input). This is the testbed used for the relocation strategies and parking management strategies


