technology_share values seem off

[jdhoffa]

See reprex. Expected values were manually calculated in excel.

---

library(r2dii.data)
library(r2dii.analysis)
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

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_ald, ~sector_ald,
  "L1",                  1000L,                           "USD",                   1000L,                            "USD",    "DL1", "direct_loantaker",     1L, "power", "company a",     "power",
  "L2",                 15000L,                           "USD",                  15000L,                            "USD",    "DL2", "direct_loantaker",     1L, "power", "company b",     "power",
  "L3",                  1200L,                           "USD",                   1200L,                            "USD",    "DL1", "direct_loantaker",     1L, "power", "company a",     "power"
)

ald <- tibble::tribble(
  ~name_company, ~sector,     ~technology, ~year, ~production, ~emission_factor, ~plant_location, ~is_ultimate_owner,
  "company a", "power",       "coalcap", 2020L,        100L,              0.5,            "US",               TRUE,
  "company a", "power",        "gascap", 2020L,        200L,              0.2,            "US",               TRUE,
  "company a", "power",      "hydrocap", 2020L,        300L,                0,            "US",               TRUE,
  "company a", "power",    "nuclearcap", 2020L,        100L,                0,            "US",               TRUE,
  "company a", "power",        "oilcap", 2020L,        100L,              0.3,            "US",               TRUE,
  "company a", "power", "renewablescap", 2020L,        200L,                0,            "US",               TRUE,
  "company b", "power",       "coalcap", 2020L,        500L,              0.5,            "US",               TRUE,
  "company b", "power",        "gascap", 2020L,          0L,              0.2,            "US",               TRUE,
  "company b", "power",      "hydrocap", 2020L,          0L,                0,            "US",               TRUE,
  "company b", "power",    "nuclearcap", 2020L,        300L,                0,            "US",               TRUE,
  "company b", "power",        "oilcap", 2020L,        100L,              0.3,            "US",               TRUE,
  "company b", "power", "renewablescap", 2020L,        100L,                0,            "US",               TRUE
)

scenario <- tibble::tribble(
  ~scenario, ~sector,     ~technology,  ~region, ~year, ~tmsr, ~smsp, ~scenario_source,
  "sds", "power",       "coalcap", "global",  2020,     1,     0,      "demo_2020",
  "sds", "power",        "gascap", "global",  2020,     1,     0,      "demo_2020",
  "sds", "power",      "hydrocap", "global",  2020,     1,     0,      "demo_2020",
  "sds", "power",    "nuclearcap", "global",  2020,     1,     0,      "demo_2020",
  "sds", "power",        "oilcap", "global",  2020,     1,     0,      "demo_2020",
  "sds", "power", "renewablescap", "global",  2020,     1,     0,      "demo_2020"
)

out <- target_market_share(
  matched, 
  ald, 
  scenario, 
  region_isos_demo
) %>% 
  filter(metric == "projected")

expected_out <- tibble::tribble(
  ~production, ~technology_share,
  448.8372093,       0.448837209,
  25.58139535,       0.025581395,
  38.37209302,       0.038372093,
  274.4186047,       0.274418605,
  100,               0.1,
  112.7906977,       0.112790698
)

out %>% 
  select(production, technology_share)
#> # A tibble: 6 x 2
#>   production technology_share
#>        <dbl>            <dbl>
#> 1      449.            0.473 
#> 2       25.6           0.0137
#> 3       38.4           0.0205
#> 4      274.            0.286 
#> 5      100             0.10  
#> 6      113.            0.107

expected_out
#> # A tibble: 6 x 2
#>   production technology_share
#>        <dbl>            <dbl>
#> 1      449.            0.449 
#> 2       25.6           0.0256
#> 3       38.4           0.0384
#> 4      274.            0.274 
#> 5      100             0.1   
#> 6      113.            0.113

^{Created on 2021-01-06 by the reprex package (v0.3.0)}

[jdhoffa]

The issue arises because there are multiple loans given to the same company. When we calculate the technology_share, it then "double counts" the companies overall production so-to-speak, and drops it's technology_share slightly lower than it should.

I will add a function earlier in the target_* function to aggregate all loans to the same company (likely should have done this a while ago).