Skip to contents

Changelog

Source: NEWS.md

humind (development version)

humind 2026.4.0

This release is a bug-fix round within the 2026 cycle that also ships the new MSNI workflow vignettes and their bundled demo datasets. It corrects the SNFI tenure-security default in add_comp_snfi(), fixes multi-column range validation in are_values_in_range(), corrects the sanitation-facility sharing counts, and tightens input validation in the education loop helpers.

Codebase Changes

New Features

  • Bundled demo datasets backing the vignette: humind_main (household-level) and the individual-level loops humind_health_ind and humind_edu_ind, documented and exposed lazily (LazyData) (#785).

Bug Fixes

  • add_loop_edu_disrupted_d_to_main(): passing attack_d = NULL (to indicate the attack dimension is absent) no longer raises object 'loop_attack' not found. The attack summary is now only built and joined when attack_d is supplied (#804).

  • add_comp_snfi(): the tenure_security_cat default now points to hlp_tenure_security, the combined security-of-tenure category produced by add_occupancy_cat() (maximum risk across occupancy and eviction risk). Previously it defaulted to hlp_occupancy_cat, so eviction risk was omitted from the SNFI composite (#801; see also #583, #584).

  • are_values_in_range(): a value outside lower/upper in any checked column now raises an error, instead of only when every column was out of range. lower/upper are also validated (both numeric, with upper >= lower), and validation messages now refer to “values” rather than “columns” (#798).

  • add_loop_edu_barrier_protection_d(): added input validation for the edu_barrier column, which is assumed to be select_one. Values must be a known barrier response code: a protection barrier (flagged 1), a non-protection/access barrier or flag code, or an undefined/non-response value (the latter two coded 0). New non_protection_issues (defaults to the non-protection edu_barrier codes) and barriers_undefined (default c("dnk", "pnta", "other")) parameters make these sets explicit; any other value (e.g. typos or combined select_multiple strings) raises an error, as do overlapping protection_issues, non_protection_issues, and barriers_undefined sets (#792).

  • add_sharing_sanitation_facility_n_ind(): the raw number of households sharing (wash_sanitation_facility_sharing_n) is no longer overwritten; the estimated number of individuals is stored in a new column (wash_sanitation_facility_sharing_n_calc). not_shared facilities no longer get an imputed individual count and leave that column as NA, so they are no longer conflated with “shared with fewer than 20 people”. The camp sanitation severity-1 classification in add_comp_wash() is now keyed on wash_sharing_sanitation_facility_cat == "not_shared" (#788).

Documentation

  • New vignette vignette("msni-workflow") (“Computing the MSNI: Humind Workflow”) providing an end-to-end, sector-by-sector walkthrough of the humind pipeline: Food Consumption (LCSI, FCS, HHS, rCSI, FCM, FCLCM), WASH (H-WISE, water quality, sanitation, hygiene), SNFI/HLP, Protection, Health, and Education (loops and main), closing with add_msni() (#785).
  • New French translation vignette("msni-workflow-fr") (“Calcul du MSNI : flux de travail Humind”) (#785).

humind 2026.3.0

This release is a patch/bug-fix round within the 2026 cycle. It tightens input validation and answer-option handling across WASH and Food Security functions, fixes inconsistent NA handling in add_hhs(), corrects the time-to-fetch-water bucketing in add_drinking_water_source_cat(), and removes the runtime dependency on impactR.utils.

Programmatic Changes

  • WASH
    • add_drinking_water_quality_jmp_cat(): the safely_managed water quality dimension deprecated in favor of ‘basic’ for improved sources with water on premises. This adheres more strictly to the JMP standards.

Codebase Changes

Enhancements

Bug Fixes

Testing

Breaking Changes

humind 2026.2.0

This release rolls out the 2026 MSNI framework update across WASH, Protection, SNFI, Food Security, and Education indicators, including a new HWISE-4 prerequisite step, a mandatory shelter-damage component, and several parameter/default changes to match this year’s Kobo form. Breaking changes are significant — read the section below before upgrading.

2026 Programmatic Changes

  • WASH
    • Water-quantity scoring is no longer computed inside add_comp_wash(). It now requires a pre-computed HWISE-4 score, produced by the new add_hwise() prerequisite function.
    • Drinking-water-quality severity thresholds shifted to match MSNI 2026 guidance.
    • The list of “unimproved” sanitation facilities was revised.
    • Default answer-option codes realigned to the updated Kobo form across several WASH functions (camp/urban/rural settings, water source, time to fetch water, handwashing facility).
  • Protection
    • The movement composite’s linear scoring formula was replaced with an explicit, capped severity scale, and gained a rule to force NA when only other_safety_measures is selected.
    • Movement options split into separate men/boys categories, with new optional weight parameters.
    • Two new movement answer options added to add_prot_score_movement(): women_girls_boys_avoid_firewood and women_girls_boys_avoid_places (both weighted 2)
    • add_prot_score_rights() gains four new answer options: yes_therapeutic_services (w=1), yes_edu_facilities (w=2), and yes_social_services (w=1) in the services sub-dimension, and yes_birth_certificates (w=1) in the justice sub-dimension. Weights for yes_healthcare and yes_schools corrected from 1 to 2. The yes_other_services (previously weight 1) and yes_other_services_justice (previously weight 1) are now unweighted and no longer contribute to the composite score.
    • add_prot_score_practices(): yes_other_activities (previously weight 2) and yes_other_social (previously weight 1) are now unweighted and no longer contribute to the composite score.
  • SNFI
    • Shelter damage is now a mandatory input to add_comp_snfi() (previously toggled on/off via a shelter_damage argument).
    • Shelter-damage category labels and severity scores were overhauled.
    • Default answer options for domestic-task assessment (add_fds_cannot_cat()) adapted to the 2026 framework. The previous logic had a mix of affirmative and negative responses resulting in FDS severity.
  • Food Security
  • Education
    • edu_ind_schooling_age_d renamed to edu_ind_age_schooling (shared default across the education loop functions).
    • child_pregnancy added as a recognized protection barrier in add_loop_edu_barrier_protection_d().
  • MSNI
    • add_msni() gains sector_in_severe_need_n and sector_severe_needs_profile outputs.

Codebase Changes

New Features

Enhancements

Bug Fixes

  • Corrected the all-missing-columns validation check in add_msni() — it previously aborted whenever any sectoral composite was missing instead of only when all of them were; calls with a partial set of composites now succeed (with a warning) instead of erroring.
  • Fixed a trailing-comma bug in add_comp_wash() (no output change).
  • Fixed NSE evaluation and modernized pipe usage in food security functions to satisfy R CMD check (internal only, no output change).

Testing

Documentation

  • Clarified that recoded HWISE items are only exposed via .keep_recoded.
  • Regenerated documentation after removing the dead sanitation_facility_jmp_cat_safely_managed parameter.

Dependency Management

Breaking Changes

  • Renames
  • WASH
    • add_comp_wash() no longer computes water-quantity scoring from raw HWISE-drink responses. The drinking_water_quantity* parameters are gone; the function now requires a pre-computed comp_wash_hwise_score column (produced by the new add_hwise()).
    • add_comp_wash(): setting_camp default changed from "camp" to c("camp_formal", "camp_informal"), and matching switched from == to %in% — default callers whose data still uses "camp" will no longer match.
    • add_comp_wash(): removed the dead sanitation_facility_jmp_cat_safely_managed parameter (the JMP category function never actually produced that value).
    • add_comp_wash(): drinking-water-quality severity thresholds shifted by one level across camp/urban/rural branches.
    • add_sanitation_facility_cat(): “unimproved” facility codes changed (dropped flush_elsewhere/plastic_bag; added twin_pit_latrine_wo_slab/other_container).
    • add_drinking_water_source_cat(): default code protected_spring → well_spring; sl_under_30_min changed from a free scalar default to a match.arg()-constrained vector (out-of-set values now error); water_on_premises default changed from "water_on_premises" to c("water_in_dwelling", "water_in_plot").
    • add_handwashing_facility_cat(): facility_undefined narrowed from c("other", "pnta") to "other"; observed/reported soap and water column-name defaults changed to _yn-suffixed names, with new value codes (e.g. yes_soap_shown → soap_available).
    • add_loop_wgq_ss(): default codes lot_of_difficulty → a_lot_of_difficulty and cannot_do → cannot_do_at_all.
  • SNFI
    • add_comp_snfi(): the shelter_damage toggle parameter was removed — shelter damage is now always required and always scored (previously optional, off by default).
    • add_shelter_damage_cat(): default damage-type codes changed (minor → minor_roof, major → major_roof, damage_windows_doors → windows_doors, damage_floors → floors, damage_walls → walls), and the returned category labels changed entirely: "No damage"/"Damaged"/"Partial collapse or destruction"/ "Total collapse or destruction"/"Undefined" → "none"/"damaged"/"part"/"total"/"undefined".
    • add_comp_snfi(): shelter-damage severity score scale shifted — total 4→5, part 3→4, damaged 2→3 (none stays 1).
    • add_fds_cannot_cat(): fds_cooking_can/fds_cooking_cannot defaults swapped (previously backwards: “can cook” defaulted to "no"); output category strings simplified ("no_cannot" → "no").
  • Protection
    • add_prot_score_movement(): the linear comp_prot_score_prot_needs_3 + 1 scoring formula was replaced by an explicit, capped scale (>= 3 ~ 4, == 2 ~ 3, == 1 ~ 2, == 0 ~ 1); the NA rule now also triggers when only other_safety_measures is selected with a zero score.
    • add_prot_score_movement(): men_boys_avoid_places split into men_avoid_places + boys_avoid_places (same split for ..._avoid_night), with new optional weight parameters (men_avoid_places_weight, men_avoid_night_weight).
    • add_prot_score_movement(): two new required columns — prot_needs_3_movement/women_girls_boys_avoid_firewood, prot_needs_3_movement/women_girls_boys_avoid_places
    • add_prot_score_rights(): four new required columns — prot_needs_1_services/yes_therapeutic_services, prot_needs_1_services/yes_edu_facilities, prot_needs_1_services/yes_social_services, prot_needs_1_justice/yes_birth_certificates

humind 2025.1.4

This release fixes a shelter classification bug affecting tents.

🛠 Codebase Changes

🛠 Bug Fixes

  • Reclassify tent as an inadequate shelter type in add_shelter_type_cat() (was incorrectly grouped with adequate shelters).

🧪 Testing

  • Updated and expanded regression tests to cover tent -> inadequate scenarios, including the priority-rule edge cases.

humind 2025.1.3

This release fixes a Protection bug related to choice-column separator handling.

🛠 Codebase Changes

🛠 Bug Fixes

🧪 Testing

  • Added unit test “the non-default separator does not affect the behavior”

humind 2025.1.2

This release introduces major terminology and logic changes across the MSNI framework, WASH, Protection, and Healthcare indicators. Several functions have new required parameters, altered defaults, or renamed outputs — breaking changes are significant.

📖 2025 Programmatic Changes

  • Terminology Migration
    All references to acute need are migrated to severe need, in line with 2025 guidance.
    • Function is_in_acute_need() is now is_in_severe_need().
    • All composite outputs renamed from *_in_acute_need → *_in_severe_need across MSNI, WASH, Health, Food Security, SNFI, Education, and Protection.
  • WASH (Handwashing Facilities)
    • Expanded logic in add_handwashing_facility_cat() to incorporate soap type (qualifying vs non-qualifying).
    • Harmonized observed vs reported classification rules.
    • Stricter handling of NA values (reported “yes” with NA → now “limited”).
  • Healthcare
  • Protection

🛠 Codebase Changes

🚀 New Features

  • WASH:
  • Protection:
    • Warnings when sub-dimension scores are NA.
    • More robust NA handling in rights and practices composites.
  • Healthcare:
    • Warning when needed == yes and received == NA.

🔄 Enhancements

  • Harmonized observed/reported classification trees for handwashing facilities.
  • Row-wise composite score computation for protection.
  • More explicit documentation of function arguments and classification logic.

🛠 Bug Fixes

  • Corrected misclassification in reported handwashing facilities where NA water/soap previously returned NA instead of “limited.”
  • Fixed inconsistent handling of vector “no” codes in handwashing classification.
  • Corrected logic in protection composites to avoid premature collapse to NA.

🧪 Testing

  • Exhaustive test grids added for handwashing classification (soap-type handling, scalar vs vector inputs).
  • Regression tests ensure new NA/warning behaviors in protection and healthcare are validated.
  • Expanded unit tests for new “severe need” naming.

📚 Documentation

  • Updated all function documentation to reflect “severe” terminology.
  • Added detailed decision tree explanation for handwashing facility classification.
  • Clarified parameter defaults for composite functions.

⚠️ Breaking Changes

  • Renames
    • is_in_acute_need() → is_in_severe_need().
    • Output columns: all *_in_acute_need → *_in_severe_need.
  • WASH
    • add_comp_wash() default parameter for drinking_water_quantity now wash_hwise_drink (was wash_drinking_water_quantity).
    • add_handwashing_facility_cat() requires new soap type columns/args and accepts vectorized “no” codes.
    • Classification rules: soap type can demote basic → limited, and NA handling has changed.
  • Protection
    • DNK/PNTA handling refined: only sub-dimensions become NA; composites NA only if both subs are NA.
  • Healthcare

humind 2025.1.1

This release introduces major updates across protection indicators, SNFI and shelter categorization, expenditure scoring, testing framework improvements, dependency management, and documentation. Deprecated WGQ components are fully removed.

📖 2025 Programmatic Changes

  • Protection: complete revamp, based on a series of new Tier 1 indicators related to Protection Needs.

  • SNFI:

    • Changes to the number of shelter issues and their mapping to the FW (from 8 to 11 total issues).
    • Inclusion of one additional indicator to the security of tenure dimension: hlp_eviction_risk.
    • Hygiene indicator removed from the FDS series - making the total number of domestic tasks (and lighting) equal to 4 instead of 5.
    • Optional shelter damages component added.

  • Health: removal of the WGQs from the framework.

  • WASH: small changes to incorporate the self-reported hygiene variants (availability of soap & water) and map them to the JMP classifications.

  • Food Security: new additional step to analyze the impact of livelihood coping strategies on Food Consumption (changes done in impactR4PHU and reflected in humind).

  • Education: small changes to indicator naming to reflect “direct attack on education”. Indicator name was edu_disrupted_occupation and is now called edu_disrupted_attack. edu_disrupted_occupation and is now called edu_disrupted_attack.

🛠 Codebase Changes

🚀 New Features

  • Protection Composite Indicators
    • Added add_comp_prot_score() as main function for computing overall Protection Composite.
    • Introduced supporting functions for rights, practices, and participation dimensions.

🔄 Enhancements

  • SNFI & Shelter Categorization
    • Revised add_shelter_issue_cat() logic for better category handling and alignment with 2025 specifications.
    • Updated add_comp_snfi() for consistent categorization and explicit handling of special cases.
  • Expenditure Indicators
    • Enhanced infrequent expenditure categorization functions and ranking logic.
  • Data Integrity
    • Composition functions now preserve original data instead of overwriting with intermediate steps.

🛠 Bug Fixes

  • Fixed improper handling of invalid inputs in multiple test cases.
  • Addressed warnings triggered by implicit handling of max() with all-NA vectors.

🧪 Testing

  • Adopted testthat best practices across test suites.
  • Expanded unit tests for:
    • Protection composite framework
    • Shelter and occupancy categorization
    • Expenditure ranking
  • Standardized warning suppression and ensured expected warnings are captured.
  • Removed obsolete tests linked to deprecated WGQ components.

📚 Documentation

  • Completed function documentation for all updated indicators.
  • Updated argument descriptions and clarified logic for SNFI and protection functions.
  • Removed all references to WGQ components.
  • Added Humind 2025 change documentation.

📦 Dependency Management

  • Removed unused packages and implicit dependency on tibble.
  • Now depends on forked impactR.utils package.

⚠️ Breaking Changes

  • Fully removed WGQ-related components and associated tests.
  • Deleted obsolete function add_child_sep_cat.R.

humind 2024.1.4

This release covers bug fixes, documentation updates, changes to parameters, and additional enhancements:

  • Documentation updates to better describe function purpose, outputs, and prerequisite functions.
  • Functions no lover use _levels but instead use separate parameters. (#547)
  • add_comp_wash() per the MSNI framework, the function was updated for the urban categorization “basic which should be score 1 instead of score 2 (#558)
  • add_comp_edu() updates to show comp_edu_score_disrupted as 3 if displaced_n is greater than 0, instead of greater than 1.(#556)
  • add_comp_health() updated so that households with at least one person with a disability have the value 3 for comp_health_score (#550)
  • add_loop_edu_ind_age_corrected() updated to allow custom starting age for education analysis based on country-specific requirements (#549)
  • add_comp_snfi() now correctly categorizes “other”” for snfi_shelter_issue_cat (#545)
  • add_loop_edu_ind_age_corrected() function error message added for date formats not in the ISO 8601 format (#513)

humind 2024.1.3

This release covers bug fixes.

  • comp_edu_score_attendance() now categorizes households correctly when children are not attending school and no protection issue was faced. (#532)
  • add_shelter_issue_cat() do not add -999 or -998 when the shelter issue is Don’t know or Prefer not to answer (it keeps NAs) and “Other” is categorized as others and not undefined. (#534)
  • add_comp_snfi() default values were fixed in documentation. (#535)
  • add_drinking_water_time_threshold_cat(): corrected input parameters “under_30min” and “above_30min”. (#539)
  • add_comp_wash(): corrected assigned severities for comp_wash_score_sanitation among HHs in camps. (#536)

humind 2024.1.2

This release covers a hot patch, a new function, and enhancements.

humind 2024.1.1

This release covers mostly hot patches on a few bugs.

  • For perceived protection risks, add_comp_prot() assigns the score of 1 to a severity level. (#514)
  • add_msni() now correctly handles metrics 3 and 4, adding a missing value when the household is not in need (#509, #510)
  • add_comp_wash has some minor refinements. (#507)
  • For total income, loans to friends and from assistance should not be considered. Function add_income_source_prop() now has these two parameters as NULL by default. It is still possible to include them. (#506)

humind 2024.1.0

Provides functions to compose usual humanitarian composite indicators related to IMPACT Multi-Sector Needs Assessments (MSNAs). Such as:

  • Food security indicators (rCSI, LCSI, HHS, FCS).
  • Washington Group Short Set of questions (WG-SS).
  • JMP ladders.
  • Specific indicators related to the MSNI framework.

The package follows the ‘Step-Composition’ approach of IMPACT R framework.

humind 0.1.900

  • Expect many breaking changes as we are moving towards a more consistent and coherent package for 2024 MSNAs
  • Food security indicators will be imported from impactR4PHU, potentially all Public Health functions.

humind 0.0.5.3

  • Bug fix: rsci() now correctly adds the rcsi_score variable.

humind 0.0.5.2

  • Breaking change: drinking_water_quantity() now recodes the frequencies of days without drinking water, while drinking_water_quantity_score() assigns a score on a 5-point scale.
  • Added: time_to_fetch_water_int_char() and time_to_fetch_water_combine() to recode and combine both variables for the time to fetch water.

humind 0.0.5.1

  • Bug fix and breaking change: handwashing_facility() now recodes the types of handwashing facility.
  • Added: handwashing_facility_score() which calculates a score on a 2-point scale based on handwashing_facility()

humind 0.0.5

  • Breaking changes: revamp of all water sources and sanitation facilities fonctions with the same approach as the food security functions.
  • Added: drinking_water_source() recodes the types of water sources, time_to_fetch_water() the time to fetch water according to a chosen threshold, and drinking_water_source_score() classify each household/individual on a 5-point scale.
  • Added: sanitation_facility() recodes the types of sanitation facilities, sharing_sanitation_facility() recodes the number of people sharing the sanitation facility, and sanitation_facility_score() classify each household on a 5-point scale.

humind 0.0.4

  • Added water_source(): Water source 5-point-scale.
  • Added sanitation_facility(): Sanitation facility 5-point scale.
  • Added handwashing_facility(): Handwashing facility 2-point scale.
  • Added drinking_water_quantity(): Drinking water quantity 5-point scale.

humind 0.0.3

  • Added hdds(): Household Dietary Diversity Scores (HDDS) (#5)
  • Added fies(): Food Insecurity Experience Scale (FIES) #6
  • Bug fix: error in lcsi() on aggregated categories’ names corrected (#20)

humind 0.0.2

  • Added: HWISE functions. (#16)
  • Updated: to dplyr v1.1.0 with the dplyr::case_when() revamp. (#)
  • Modified: dependency know goes to impactR.utils and not impactR. (#)

humind 0.0.1

  • A few patches to github-action and pages

humind 0.0.0.9000

  • Initial commit