8.3.9. Correlation Demand Files

It is possible to use macrodrivers as a way to infer the service demand. For example, one can use the expected GDP based on purchasing power parity (GDP PPP) and population in the future per region to infer the service demand using a regressor.

To do this, a minimum of three files are required:

1. A macrodrivers file

2. A file which states the regression parameters

3. A file which dictates how the demand per benchmark year is split across the timeslices.

These files (explained in more detail below) should be referenced from the TOML settings file using the macrodrivers_path, regression_path, and timeslice_shares_path keys respectively.

8.3.9.1. Macrodrivers

An example of a shortened macrodriver file is shown below. This file contains the data for each of the years you are interested in. For example, in the file below, it contains GDP PPP in region R1, in the unit millionUS$2015 for each year. It also contains the data for the population.

Macrodrivers

variable	region	unit	2010	2011	…
GDP|PPP	R1	millionUS$2015	1206919	1220599	…
Population	R1	million	80.0042	81.82599	…

variable

This is the variable that you would like to use in the regression for the service demand.

region

This is the region that the data applies to. This must correlate with the regions set in the rest of your input files, as well as the toml file.

unit

This unit can be whatever you like, however they must be consistent across all input files.

Years (one column per year)

This is the quantity of the variable per year of the simulation.

8.3.9.2. Regression Parameters

In the regression parameters file, it is necessary to state the parameters of the regression. This can be obtained from your own dataset, where you regress the service demand against GDP PPP and populaiton, for example.

An example file is shown below:

Regression Parameters File

sector	function_type	coeff	region	electricity	gas	heat	CO2f
Residential	logistic-sigmoid	GDPexp	R1	0	0	9.94E-02	0
Residential	logistic-sigmoid	constant	R1	0	0	0.0000434	0
Residential	logistic-sigmoid	GDPscaleLess	R1	0	0	753.1068725	0
Residential	logistic-sigmoid	GDPscaleGreater	R1	0	0	672.9316672	0

sector

This is the sector name in which these parameters apply to.

function_type

This is the function type you would like to MUSE to use. MUSE allows these to be:

• Exponential

• ExponentialAdj

• Logistic

• Loglog

• LogisticSigmoid

• Linear

• endogenous_demand

Your own functions can be created using the @register_regression hook, from the regressions.py file.

coeff

This is the coefficient for the respective function type. These are explicitly defined within the regressions.py file, as they differ between functions.

region

This is the region in which these parameters apply to.

Commodities (one column per commodity)

Here you can specify the coefficients for the expected demand for the respective commodity.

8.3.9.3. Timeslice share

In this file, you are able to split the energy service proportionally by timeslice.

An example file is shown below:

Timeslice share

timeslice	region	electricity	gas	heat	CO2f	wind
1	R1	0	0	0.034835	0	0
2	R1	0	0	0.064546	0	0
3	R1	0	0	0.044569	0	0
4	R1	0	0	0.011161	0	0
5	R1	0	0	0.014145	0	0
6	R1	0	0	0.085783	0	0

timeslice

This is the timeslice index.

region

This is the region in question for this data.

Commodities (one column per commodity)

Here you specify the proportion of each energy service for each timeslice.