data-raw/README.md

In JGCRI/eiamod: Economic Indicator Modeling using EIA Data

Raw data processing for the package

Dataset outputs

This directory and its subdirs contain the code and data needed to produce the data files used by the modeling functions in the package. All of the outputs from these functions are saved as package data and documented in the package documentation (accessed with help(dataset-name) from within R).

Datasets included: region_hourly : load and temperature by region and hour residual_load : residual load by region and hour energy_weekly : weekly petroleum and gas consumption gdp_quarterly : quarterly national GDP values

Code and processing

Petroleum, gas, and GDP data

Raw data for these variables, as produced by the eiafcst python package, is stored in the gdp subdir. The data is subdivided into dev, train, and test subsets, but these distinctions are not used in this package. The data is converted from its original python save format to csv by the code in gdp/dev/fixdata.ipynb. Although this code lives in the dev subdir, it processes all three datasets. The csv files are processed by the code in gdp/procgdp.R and saved in the package datasets listed above.

Electricity and temperature data

Electricity and temperature data are processed using the raw data in temperature_by_agg_region.csv.gz and load_by_agg_region.csv.gz, both of which are produced by the eiafcst python package. These data are read in and saved as package data by the code in temp_elec.R.

The code in residual_load.R fits a MARS model for load from temperature and subtracts the predictions from the actual load to produce a residual load. This program makes use of the region_hourly dataset, so that data must be produced and the eiamod package rebuilt before this code can be run.

JGCRI/eiamod documentation built on Feb. 21, 2020, 9:16 a.m.