scotgovanalysis / vedar Goto Github PK

View Code? Open in Web Editor NEW

5.0 9.0 1.0 540 KB

A collection of functions to enable the import and analysis of the outputs of Veda TIMES runs in R.

R 100.00%

vedar's Introduction

vedar

A collection of functions to enable the import and analysis of the outputs of Veda TIMES runs in R.

Installation

Install vedar from GitHub with:

# install.packages("devtools")
devtools::install_github("datasciencescotland/vedar", build_vignettes = T)

If the above does not work, you can install from source:

Go to the vedar repository on GitHub
Click Clone or download then Download ZIP
Save the file locally and Unzip
Install with install.packages()

install.packages("your/directory/vedar", repos = NULL, type="source")

Usage

import_vd("vd_filename.VD") imports the VD data to a tibble
import_vds("vd_filename.VDS") imports the VDS set information to tibble
import_vde("vd_filename.VDE") imports the VDE description information to tibble
import_vdt("vd_filename.VDT") imports the structural data that is available to the model, as well as user constraints. The latter is indicated by direction == "uc".
prep_data("vd_filename_base") calls the three import_* functions, joins data, and standardises case and missing data.
prep_sector_dat(sector_dat) converts cases to lower and removes spaces from column names in sector_dat. sector_dat needs to exist as a tibble. Can be imported with read_*()
prep_vdt_data calls import_vdt() and import_vde joins the data, standardises case and missing data.
define_sector_from_list(dat, join_variable_name, sector_dat, sector_info_column, sector_dat_join_variable_colsector_dat) appends the sector information in sector_dat to dat joining on join_variable_name. See ?define_sector_from_list
make_res(dat, node_labels, edge_labels, sankey_width, sankey_height, font_size, use_weights, input_data_type) creates a RES from vd or vdt tibble. Data must only contain a single region. Magnitudes of flows are represented if data includes only a single period, and if use_weights = F.
make_graph_from_veda_df(dat, node_labels, edge_labels, input_data_type) creates an igraph graph object from a vd or vdt tibble. If only a single period in vd data, the weights of the edges are set to the value of a var_fout of the commodity. Data must only contain single region in case process names are repeated across regions.
make_res_from_graph(g, edge_labels, sankey_width, sankey_height, font_size) creates a RES from an igraph object created by make_graph_from_veda_df, or a subgraph. Sankey edge widths represent the edge_weights.
check_in_path(node_regex, path): Check whether a process (graph node) containing a specified regular expression is present in a list of paths
syscost(dat) extract the system cost by region

vedar's People

Contributors

Stargazers

Watchers

Forkers

ramap27

vedar's Issues

make_res has no default font size

https://github.com/DataScienceScotland/vedar/blob/0d45a8155c3710d9cc2f9a85879bc1ce2974f560/R/network_analysis.R#L34

Not a big issue, obviously, but having a default would give at least a starting point for anyone having a go.

data check - variable that changes by more than user specified threshold

fix make_graph_from_veda_df to account for region

processes, commodities can be repeated across regions. Return error for graph over more than one region or create separate graphs for each region

fix make_graph_from_veda_df for R4.1.1

use of summarise in computation of edge weights causes an issue when values of non-grouped variables are distinct. Replace summarise call with left_join()

deselect region in sets file

sets file deselects region. Include this to allow for multiregional models and join by region and variable

data check - commodity not passing through expected node

add specify dplyr:: for nodes <- in make_res

period and vintage digits dropped

add correct demos_001_sector

demos_001_sector includes "-" and "none". The make file needs to be rerun

altered fieldsize specification in vd file header

Fieldsize specifcation if vd file header has changed from 31 to 63. vd_structure_match_expected() fails

add specification of grouping and summarising edge values in make_res()

add roxygen2 details to get_veda_dat.R

fix to allow NULL options for make_res()

Update make_res Roxygen2 example to use correct arguments

vignette remove repeat of filename_base

add make_graph_from_veda_df to vignette

Add all datasets to Data.R

DemoS_001 is repeated, and most others are missing

pass sector column name to make_res

make define_sector_from_list arguments user friendly

fix missing dplyr:: spec in make_res and test_make_graph_from_veda_df

Check make_graph_from_veda_df for missing descriptions

values_fn warning/error in pivot_wider

call to pivot_wider in make_edges returns a warning in old versions of tidyr, and error in newer versions. The values_fn function needs to be specified, but doing this with tidyeval is non-trivial

Asked on stack
https://stackoverflow.com/questions/71423206/using-pivot-wider-in-a-function-with-named-variable-from-function-argument

change make_res to enable filtering by non sector grouping variables

The make_res function will need to be changed, and the relevant section in the vignette

fix make_graph_from_veda_df generation of duplicate rows

make_graph_from_veda_df assigns duplicate rows from

edges <- edges %>%
      dplyr::left_join(dat %>%
                       dplyr::filter(attribute == "var_fout") %>%
                       dplyr::select(commodity, commodity_description, pv) %>%
                       unique()

Need to join on process/source and possibly sum over timeslice

Fix the plot render size in the vignette to be consistent

Plot rendering size differs on different machines when vignette is built. The sankey_height and sankey_width options control this, but appear to work differently on different machines. So this needs to be investigated and fixed

create set information from snt

make_res requires dat to have a sector column

https://github.com/DataScienceScotland/vedar/blob/0d45a8155c3710d9cc2f9a85879bc1ce2974f560/R/network_analysis.R#L45

 if(sector_select %in% dat$sector == F){

The column name for the sectors is hardcoded here - whereas in my messing around I have a process_sector and a commodity_sector, much like the example output of define_sector_from_list which has the column name process_major_sector.

There are also many cases where process_sector and commodity_sector don't match, so not easy to know which to pick. Probably process_sector, as the Sankey diagram has the commodities as flows between processes?

sector_dat %>%
#strings to lower
dplyr::mutate_if(is.character, stringr::str_to_lower)

to:

sector_dat <- sector_dat %>%
#strings to lower
dplyr::mutate_if(is.character, stringr::str_to_lower)

Otherwise the mutate output is not being stored anywhere?

(Sorry if this is in the wrong place, I don't know how to use GitHub!)