davecook1985/home-care-cost-model

---
license: cc-by-4.0
task_categories:
- tabular-regression
- tabular-classification
language:
- en
- fr
tags:
- home-care
- canada
- health-policy
- cost-model
- aging-in-place
- personal-support-worker
- tax-credit
size_categories:
- 1K<n<10K
pretty_name: Home Care Cost Model — Canadian Service Mix, Rate Bands, and Tax Relief Sensitivity
configs:
- config_name: services_canada
data_files: home_care_services_canada.csv
- config_name: tax_parameters_2026
data_files: home_care_tax_parameters_2026.csv
- config_name: subsidy_programs
data_files: home_care_subsidy_programs.csv
- config_name: scenarios
data_files: home_care_scenarios.csv
- config_name: per_province_rate_bands
data_files: home_care_per_province_rate_bands.csv
- config_name: cost_model_archetypes
data_files: home_care_cost_model_archetypes.csv
- config_name: tax_relief_sensitivity
data_files: home_care_tax_relief_sensitivity.csv
- config_name: subsidy_gap
data_files: home_care_subsidy_gap.csv
---

# Home Care Cost Model — Datasets

Open data companion to the Home Care Cost Model reference framework. Eight
CSV datasets totalling roughly 9,200 rows span three classes: raw source
retrievals, hand-curated reference tables, and engine-derived scenario
grids. All values are denominated in Canadian dollars (CAD) and indexed to
the 2026 taxation year.

## Dataset classes

### Class 1 — Sources (raw, credited)

Stored under `sources/<organisation>/`. Every file has a sibling
`<filename>.source.json` recording upstream URL, retrieval ISO timestamp,
declared license, and SHA256. A machine-readable provenance manifest is at
`SOURCES.md`. No Source file is ever edited post-download. See
`pull_sources.py` to reproduce.

Upstreams include Statistics Canada (wage, population, and CPI tables),
Canadian Institute for Health Information (home care indicators), Canada
Revenue Agency (tax folios), Veterans Affairs Canada, Indigenous Services
Canada (FNIHB), and the Ontario, BC, Alberta, Quebec, Saskatchewan,
Manitoba, Nova Scotia, New Brunswick, Prince Edward Island, Newfoundland
and Labrador, Yukon, Northwest Territories, and Nunavut ministries of
health.

### Class 2 — Reference tables (hand-curated, deterministic)

| File | Rows | Description |
|---|---|---|
| `home_care_services_canada.csv` | ~80 | Scope of practice, training hours, regulating body, and median private-pay rate bands for 8 service categories across 13 Canadian jurisdictions plus federal VAC VIP and ISC FNIHB cross-jurisdiction rows. |
| `home_care_tax_parameters_2026.csv` | ~65 | 2026 federal and provincial tax parameters for METC, DTC, CCC, VAC VIP, and provincial disability amounts. |
| `home_care_subsidy_programs.csv` | 15 | Provincial, territorial, VAC VIP, and FNIHB subsidised home care programs: administering body, eligibility, typical hours, wait time, URL. |

Each row in these tables cites one or more `source_id`s from the Sources
manifest. Rates were reconciled against StatsCan Table 14-10-0417-01 wages
as a floor and the Canadian Home Care Association agency markup range as a
ceiling.

### Class 3 — Derived scenarios (engine-generated)

Every row in every derived dataset is produced by importing
`engines/python/engine.py` and calling `calculate_home_care_costs()` with
a specific input combination. These are the tables most directly useful
for secondary analysis.

| File | Rows | Description |
|---|---|---|
| `home_care_scenarios.csv` | 5,000 | Synthetic household scenarios. Province sampled by StatsCan 65+ population share. Diagnosis-cognition-mobility joint distribution calibrated to CIHI home care indicators. Deterministic with `random.seed(42)`. |
| `home_care_per_province_rate_bands.csv` | 520 | Per-province, per-category, per-year (2019–2026) rate bands, CPI-adjusted from the 2026 reference using the StatsCan 18-10-0004-01 health subindex. |
| `home_care_cost_model_archetypes.csv` | 1,820 | Engine outputs for a (province × ADL × IADL × cognitive × mobility) grid at constant household composition and income. |
| `home_care_tax_relief_sensitivity.csv` | 1,300 | Sensitivity grid: 13 provinces × 5 ADL points × 5 income bands × 4 DTC/VIP flag combinations. |
| `home_care_subsidy_gap.csv` | 390 | 13 jurisdictions × 30 representative scenarios, showing subsidised hours vs model-recommended hours vs monthly dollar gap. Exposes cross-province home care program inequity. |

## "Synthetic. Generated from seed 42. Not survey data."

All Class 3 CSVs are produced deterministically from calibrated sampling
distributions and the reference tables. They are **not** survey data and
**must not** be cited as empirical observations of individual households.
They are intended for policy analysis, decision support, and cost model
validation, not for biostatistical inference.

## Reproducing the pipeline

```bash
# From the project root
python datasets/pull_sources.py # Class 1
python datasets/generate_home_care_services_canada.py # Class 2a
python datasets/generate_home_care_tax_parameters.py # Class 2b
python datasets/generate_home_care_subsidy_programs.py # Class 2c
python datasets/generate_home_care_scenarios.py # Class 3a
python datasets/generate_home_care_per_province_rate_bands.py # Class 3b
python datasets/generate_home_care_cost_model_archetypes.py # Class 3c
python datasets/generate_home_care_tax_relief_sensitivity.py # Class 3d
python datasets/generate_home_care_subsidy_gap.py # Class 3e
```

Regenerating any generator twice in the same Python environment must
yield a byte-identical output — all sampling uses `random.seed(42)`.

## License

- **Source code** (`generate_*.py`, `pull_sources.py`): MIT
- **Reference and derived CSVs**: Creative Commons Attribution 4.0
International (CC BY 4.0). Attribution: *Dave Cook (2026). Home Care
Cost Model — datasets companion to the working paper.*

## Citation

```
Cook, D. (2026). The Home Care Cost Model: Personal Support,
Housekeeping, and Service Mix Decisions for Aging in Place in Canada
— datasets. Version 0.1.0. https://doi.org/10.5281/zenodo.19491364
```

The canonical citable identifier is the Zenodo DOI
[10.5281/zenodo.19491364](https://doi.org/10.5281/zenodo.19491364).

## Related resources

This dataset is one facet of a larger open release. The same data, code,
and working paper are mirrored across the following canonical locations:

- **Working paper (PDF):** <https://www.binx.ca/guides/home-care-cost-model-guide.pdf>
- **Source code (GitHub):** <https://github.com/DaveCookVectorLabs/home-care-cost-model>
- **Archival DOI (Zenodo):** <https://doi.org/10.5281/zenodo.19491364>
- **Kaggle mirror:** <https://www.kaggle.com/datasets/davecook1985/home-care-cost-model>
- **Internet Archive:** <https://archive.org/details/home-care-cost-model-binx>
- **OSF project:** <https://osf.io/2fc4a/>
- **Wikidata entity:** [Q139082584](https://www.wikidata.org/wiki/Q139082584)
- **Python engine (PyPI):** <https://pypi.org/project/home-care-cost-model/>
- **JavaScript engine (npm):** <https://www.npmjs.com/package/@davecook/home-care-cost-model>
- **Rust engine (Crates.io):** <https://crates.io/crates/home-care-cost-model-engine>
- **Documentation (ReadTheDocs):** <https://home-care-cost-model.readthedocs.io/>

---

Maintained by Dave Cook of [Binx Professional Cleaning](https://www.binx.ca), North Bay and Sudbury, Ontario.

许可证：CC BY 4.0（知识共享署名4.0国际许可协议）
任务类别：表格回归（tabular-regression）、表格分类（tabular-classification）
语言：英语（en）、法语（fr）
标签：家庭护理（home-care）、加拿大（canada）、卫生政策（health-policy）、成本模型（cost-model）、原地养老（aging-in-place）、个人护理员（personal-support-worker）、税收抵免（tax-credit）
数据规模：1000条 < 数据量 < 10000条
展示名称：家庭护理成本模型——加拿大服务组合、费率区间与税收减免敏感性分析
配置项：
- 配置名：services_canada，数据文件：home_care_services_canada.csv
- 配置名：tax_parameters_2026，数据文件：home_care_tax_parameters_2026.csv
- 配置名：subsidy_programs，数据文件：home_care_subsidy_programs.csv
- 配置名：scenarios，数据文件：home_care_scenarios.csv
- 配置名：per_province_rate_bands，数据文件：home_care_per_province_rate_bands.csv
- 配置名：cost_model_archetypes，数据文件：home_care_cost_model_archetypes.csv
- 配置名：tax_relief_sensitivity，数据文件：home_care_tax_relief_sensitivity.csv
- 配置名：subsidy_gap，数据文件：home_care_subsidy_gap.csv

# 家庭护理成本模型——数据集

本数据集为《家庭护理成本模型》参考框架的配套开放数据。共包含8个CSV数据集，总计约9200条数据，分为三类：原始来源检索数据、人工整理参考表、引擎生成的情景网格。所有数值均以加拿大元（CAD）计价，并以2026纳税年度为基准索引。

## 数据集类别

### 类别1：来源数据（原始、带溯源）
数据存储于`sources/<组织机构>/`路径下。每个数据文件均配有同名的`<filename>.source.json`附属文件，记录上游来源URL、检索的ISO时间戳、声明的许可证及SHA256哈希值。`SOURCES.md`中包含机器可读的溯源清单。所有来源文件在下载后均不进行任何编辑。可通过`pull_sources.py`脚本复现数据获取流程。

上游数据来源包括：加拿大统计局（薪资、人口及消费者物价指数表格）、加拿大卫生信息研究所（家庭护理指标）、加拿大税务局（税务手册）、加拿大退伍军人事务部、加拿大原住民服务部（First Nations and Inuit Health Branch, FNIHB，原住民健康与福利分局），以及安大略省、不列颠哥伦比亚省、阿尔伯塔省、魁北克省、萨斯喀彻温省、曼尼托巴省、新斯科舍省、新不伦瑞克省、爱德华王子岛省、纽芬兰与拉布拉多省、育空地区、西北地区和努纳武特地区的卫生部门。

### 类别2：参考表（人工整理、确定性）

| 文件名 | 数据量 | 说明 |
|---|---|---|
| `home_care_services_canada.csv` | ~80 | 涵盖加拿大13个司法管辖区，以及联邦退伍军人事务部VIP项目和原住民健康与福利分局的跨辖区数据，包含8类服务的执业范围、培训时长、监管机构及私人付费中位费率区间。 |
| `home_care_tax_parameters_2026.csv` | ~65 | 2026年联邦及各省的税收参数，涵盖医疗费用税收抵免（Medical Expense Tax Credit, METC）、残障税收抵免（Disability Tax Credit, DTC）、加拿大儿童福利（Canada Child Benefit, CCC）、退伍军人事务部VIP项目及省级残障补助额度。 |
| `home_care_subsidy_programs.csv` | 15 | 各省、地区、退伍军人事务部VIP项目及原住民健康与福利分局的补贴型家庭护理项目信息，包含管理机构、申领资格、典型服务时长、等待时间及相关URL。 |

上述参考表中的每一行数据均引用了来源清单中的一个或多个`source_id`。费率区间以加拿大统计局表格14-10-0417-01中的薪资作为下限，以加拿大家庭护理协会的机构加价范围作为上限进行校准。

### 类别3：衍生情景数据（引擎生成）
所有衍生数据集中的每一行数据，均通过导入`engines/python/engine.py`并调用`calculate_home_care_costs()`函数，结合特定输入组合生成。此类表格是最适用于二次分析的数据集。

| 文件名 | 数据量 | 说明 |
|---|---|---|
| `home_care_scenarios.csv` | 5,000 | 合成家庭情景数据集。按照加拿大统计局65岁以上人口占比抽样省份，结合加拿大卫生信息研究所的家庭护理指标校准诊断-认知-移动能力联合分布。采用`random.seed(42)`确保结果可复现。 |
| `home_care_per_province_rate_bands.csv` | 520 | 各省、每类服务、各年度（2019–2026）的费率区间，以2026年为基准，通过加拿大统计局18-10-0004-01健康分项消费者物价指数进行通胀调整。 |
| `home_care_cost_model_archetypes.csv` | 1,820 | 基于（省份×日常生活活动（Activities of Daily Living, ADL）×工具性日常生活活动（Instrumental Activities of Daily Living, IADL）×认知能力×移动能力）的网格引擎输出结果，保持家庭构成与收入水平恒定。 |
| `home_care_tax_relief_sensitivity.csv` | 1,300 | 敏感性分析网格：13个省份×5个日常生活活动评分点×5个收入档位×4个残障税收抵免（Disability Tax Credit, DTC）/VIP项目标识组合。 |
| `home_care_subsidy_gap.csv` | 390 | 13个司法管辖区×30个代表性情景，展示补贴服务时长、模型推荐服务时长与月度美元差额，揭示跨省份家庭护理项目的公平性差异。 |

> 「合成数据。基于种子42生成，非调查数据。」

所有类别3的CSV数据集均通过校准后的抽样分布与参考表确定性生成。本数据集**绝非调查数据**，**不得作为单个家庭的实证观测结果引用**。其设计用途为政策分析、决策支持与成本模型验证，而非用于生物统计推断。

## 复现数据流程

bash
# 从项目根目录执行
python datasets/pull_sources.py # 生成类别1数据
python datasets/generate_home_care_services_canada.py # 生成类别2a数据
python datasets/generate_home_care_tax_parameters.py # 生成类别2b数据
python datasets/generate_home_care_subsidy_programs.py # 生成类别2c数据
python datasets/generate_home_care_scenarios.py # 生成类别3a数据
python datasets/generate_home_care_per_province_rate_bands.py # 生成类别3b数据
python datasets/generate_home_care_cost_model_archetypes.py # 生成类别3c数据
python datasets/generate_home_care_tax_relief_sensitivity.py # 生成类别3d数据
python datasets/generate_home_care_subsidy_gap.py # 生成类别3e数据

在同一Python环境中重复运行任意生成脚本，将得到字节完全一致的输出——所有抽样均使用`random.seed(42)`固定随机种子。

## 许可证

- **源代码**（`generate_*.py`、`pull_sources.py`）：MIT许可证
- **参考表与衍生CSV数据集**：知识共享署名4.0国际许可协议（CC BY 4.0）。署名要求：*戴夫·库克（2026）. 家庭护理成本模型——配套工作论文的数据集.*

## 引用格式

Cook, D. (2026). The Home Care Cost Model: Personal Support,
Housekeeping, and Service Mix Decisions for Aging in Place in Canada
— datasets. Version 0.1.0. https://doi.org/10.5281/zenodo.19491364

库克, D. (2026). 家庭护理成本模型：加拿大原地养老的个人护理、家政服务与服务组合决策——数据集. 版本0.1.0. https://doi.org/10.5281/zenodo.19491364

标准可引用标识符为Zenodo DOI [10.5281/zenodo.19491364](https://doi.org/10.5281/zenodo.19491364)。

## 相关资源

本数据集为大型开放发布项目的组成部分。以下为数据、代码与工作论文的官方镜像站点：
- **工作论文（PDF）**：<https://www.binx.ca/guides/home-care-cost-model-guide.pdf>
- **源代码（GitHub）**：<https://github.com/DaveCookVectorLabs/home-care-cost-model>
- **存档DOI（Zenodo）**：<https://doi.org/10.5281/zenodo.19491364>
- **Kaggle镜像**：<https://www.kaggle.com/datasets/davecook1985/home-care-cost-model>
- **互联网档案馆镜像**：<https://archive.org/details/home-care-cost-model-binx>
- **开放科学框架（OSF）项目**：<https://osf.io/2fc4a/>
- **维基数据实体**：[Q139082584](https://www.wikidata.org/wiki/Q139082584)
- **Python引擎（PyPI）**：<https://pypi.org/project/home-care-cost-model/>
- **JavaScript引擎（npm）**：<https://www.npmjs.com/package/@davecook/home-care-cost-model>
- **Rust引擎（Crates.io）**：<https://crates.io/crates/home-care-cost-model-engine>
- **文档（ReadTheDocs）**：<https://home-care-cost-model.readthedocs.io/>

---

本数据集由安大略省北湾与萨德伯里市的[Binx专业清洁公司](https://www.binx.ca)戴夫·库克维护。