SDA: first year of scenario should determine start of analysis

[jacobvjk]

r2dii.analysis/R/target_sda.R

Line 366 in cd3611c

right_join(

For the target_sda() function the linked line shows that the scenario data is right joined onto the abcd. This means that in case the abcd has values starting in a year earlier than the first scenario year, this year will not be dropped from the analysis and the scenario thus does not determine the start year. Applying an inner join should fix this.

For comparison, in target_market_share() the following join is used and thus the first year of the scenario data determines the start year:

r2dii.analysis/R/join_abcd_scenario.R

Line 65 in cd3611c

inner_join(scenario, by = scenario_columns()) %>%

Reprex:

library(tidyverse)
library(r2dii.data)
library(r2dii.match)
library(r2dii.analysis)

lbk_raw <- r2dii.data::loanbook_demo

# let us have the production data start a year befor the initial years in the scenario files
abcd <- r2dii.data::abcd_demo %>% 
  filter(year >= 2019)

matched <- r2dii.match::match_name(lbk_raw, abcd = abcd) %>% 
  r2dii.match::prioritize()

# min year for tmsr/smsp and sda sceanrios is the same
r2dii.data::scenario_demo_2020 %>% 
  pull(year) %>% 
  min()
#> [1] 2020

r2dii.data::co2_intensity_scenario_demo %>% 
  pull(year) %>% 
  min()
#> [1] 2020

# calculation for EI sectors, using SDA  
result_sda <- matched %>% 
  r2dii.analysis::target_sda(
    abcd = abcd,
    co2_intensity_scenario = r2dii.data::co2_intensity_scenario_demo,
    region_isos = r2dii.data::region_isos_demo
  )
#> Warning: Removing rows in abcd where `emission_factor` is NA

# calculation for production pathway sectors, using target market share
result_market_share <- matched %>% 
  r2dii.analysis::target_market_share(
    abcd = abcd,
    scenario = r2dii.data::scenario_demo_2020,
    region_isos = r2dii.data::region_isos_demo
  )

# comparison of intial years in results
result_sda %>% 
  pull(year) %>% 
  min()
#> [1] 2019

result_market_share %>% 
  pull(year) %>% 
  min()
#> [1] 2020
Created on 2023-07-19 with [reprex v2.0.2](https://reprex.tidyverse.org/)

[jdhoffa]

Seems like it may be a little more complicated than just changing that line, but i am having a look now.

[jdhoffa]

Ok actually I don't think this requires any action.

Notice that in the following reprex, all target values start at the start of the scenario, while all "measured" values (ie. corporate economy and projected values) begin at the first value available. To me this is expected behaviour:

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(r2dii.data)
library(r2dii.analysis)

matched <- tibble::tribble(
  ~id_loan, ~loan_size_outstanding, ~loan_size_outstanding_currency, ~loan_size_credit_limit, ~loan_size_credit_limit_currency, ~id_2dii,            ~level, ~score,      ~sector,     ~name_abcd, ~sector_abcd,
  "L162",                      1,                           "EUR",                       2,                            "EUR",    "UP1", "ultimate_parent",      1, "automotive", "shaanxi auto",     "cement"
)

abcd <- tibble::tribble(
  ~name_company,  ~sector, ~technology, ~year, ~production, ~emission_factor, ~plant_location, ~is_ultimate_owner,
  "shaanxi auto", "cement",       "ice",  2024,           1,                1,            "BF",               TRUE,
  "shaanxi auto", "cement",       "ice",  2025,           1,                1,            "BF",               TRUE
)

scenario <- tibble::tribble(
  ~scenario,  ~sector,  ~region, ~year, ~emission_factor,           ~emission_factor_unit, ~scenario_source,
  "b2ds", "cement", "global",  2025,              0.5, "tons of CO2 per ton of cement",      "demo_2020",
  "b2ds", "cement", "global",  2050,              0.2, "tons of CO2 per ton of cement",      "demo_2020"
)

out <- target_sda(
  matched,
  abcd,
  scenario,
  region_isos = r2dii.analysis:::region_isos_stable
)

out_actual_data <- filter(out, !grepl("target", emission_factor_metric))
min(out_actual_data$year)
#> [1] 2024

out_targets <- filter(out, grepl("target", emission_factor_metric))
min(out_targets$year)
#> [1] 2025

^{Created on 2023-07-19 with reprex v2.0.2}

[jdhoffa]

In any case, I have opened #440 to prove in a test that the expected behavior is maintained.