Update README.md

README.md

@@ -11,63 +11,207 @@ pretty_name: hopecorner
 size_categories:
 - 10K<n<100K
 configs:
-  - config_name: bicycle_repairs
-    data_files:
-      - split: train
-        path: data/bicycle_repairs_rows.csv
-  - config_name: blocked_slots
-    data_files:
-      - split: train
-        path: data/blocked_slots_rows.csv
-  - config_name: daily_notes
-    data_files:
-      - split: train
-        path: data/daily_notes_rows.csv
-  - config_name: donations
-    data_files:
-      - split: train
-        path: data/donations_rows.csv
-  - config_name: guest_proxies
-    data_files:
-      - split: train
-        path: data/guest_proxies_rows.csv
-  - config_name: guests
-    data_files:
-      - split: train
-        path: data/guests_rows.csv
-  - config_name: haircut_visits
-    data_files:
-      - split: train
-        path: data/haircut_visits_rows.csv
-  - config_name: laundry_bookings
-    data_files:
-      - split: train
-        path: data/laundry_bookings_rows.csv
-  - config_name: meal_attendance
-    data_files:
-      - split: train
-        path: data/meal_attendance_rows.csv
-  - config_name: shower_reservations
-    data_files:
-      - split: train
-        path: data/shower_reservations_rows.csv
+- config_name: bicycle_repairs
+  data_files:
+  - split: train
+    path: data/bicycle_repairs_rows.csv
+- config_name: blocked_slots
+  data_files:
+  - split: train
+    path: data/blocked_slots_rows.csv
+- config_name: daily_notes
+  data_files:
+  - split: train
+    path: data/daily_notes_rows.csv
+- config_name: donations
+  data_files:
+  - split: train
+    path: data/donations_rows.csv
+- config_name: guest_proxies
+  data_files:
+  - split: train
+    path: data/guest_proxies_rows.csv
+- config_name: guests
+  data_files:
+  - split: train
+    path: data/guests_rows.csv
+- config_name: haircut_visits
+  data_files:
+  - split: train
+    path: data/haircut_visits_rows.csv
+- config_name: laundry_bookings
+  data_files:
+  - split: train
+    path: data/laundry_bookings_rows.csv
+- config_name: meal_attendance
+  data_files:
+  - split: train
+    path: data/meal_attendance_rows.csv
+- config_name: shower_reservations
+  data_files:
+  - split: train
+    path: data/shower_reservations_rows.csv
 ---
 
-# About
+# Dataset Card for Hope's Corner
 
-This repository contains anonymized operational data from Hope's Corner, a non-profit organization in Mountain View, CA, providing meals, laundry, showers, haircuts, and bicycle repair services to unhoused and low-income community members.
+Anonymized operational data from [Hope's Corner](https://hopescorner.org/), a non-profit in Mountain View, CA that provides meals, laundry, showers, haircuts, and bicycle repair to unhoused and low-income community members. The dataset covers guest visits and service records across ten relational tables, linked by anonymized guest IDs.
 
-## Problem Statements
+## Dataset Details
 
-### Predicting Daily Meal Demand
-Forecast daily meal demand using historical check-in records and explore external predictors such as weather impacts (rain, extreme temperatures) and calendar events (federal holidays).
+### Dataset Description
 
-### Identifying Service Gaps and Equity in Guest Support
-Analyze demographic and housing status differences in service utilization, identify service gaps, and examine potential inequities in support access (e.g., waiting list times or representation across programs).
+Hope's Corner runs drop-in services for community members who are unhoused or low-income. Staff log each service interaction in an internal system, recording which guests attended, what services they used, and operational notes. This dataset is an anonymized export of those records, structured as relational CSV files. It covers meal attendance, laundry and shower bookings, haircut visits, bicycle repairs, donations received, and guest demographic information.
 
-## Data Schema
+The two main research questions this dataset is designed to support are forecasting daily meal demand and identifying gaps or inequities in service access across different guest populations.
 
-The dataset consists of relational CSV files in the `data/` directory, linked by anonymized guest IDs.
+- **Curated by:** Hope's Corner, Mountain View, CA
+- **Language(s) (NLP):** English
+- **License:** MIT
+
+### Dataset Sources
+
+- **Source:** Internal operational records from Hope's Corner
+- **Repository:** `data/` directory, one CSV per table
+
+## Uses
+
+### Direct Use
+
+- **Meal demand forecasting:** Predict daily attendance using historical check-in records, and explore external predictors like weather (rain, extreme temperatures) and calendar events (federal holidays, local events).
+- **Service gap analysis:** Identify which guest populations use which services, spot underserved groups, and examine differences in wait times or access across demographics and housing statuses.
+- **Operational planning:** Understand slot utilization for laundry, showers, and haircuts to inform scheduling decisions.
+- **Donation tracking:** Analyze the volume and type of donations over time relative to service demand.
+
+### Out-of-Scope Use
+
+- Should not be used to re-identify individual guests. Records are anonymized and must remain so.
+- Not suitable for any use that could be used to surveil, profile, or target unhoused individuals.
+- Not a representative sample of all unhoused individuals in Mountain View or the broader region; it reflects only guests who visited Hope's Corner.
+
+## Dataset Structure
+
+### Tables
+
+The dataset is organized as ten relational CSV files in the `data/` directory, all linked by anonymized guest UUIDs.
+
+| Table | File | Description |
+|---|---|---|
+| `guests` | `guests_rows.csv` | One row per guest; core demographic and status fields |
+| `guest_proxies` | `guest_proxies_rows.csv` | Records when one guest picks up services on behalf of another |
+| `meal_attendance` | `meal_attendance_rows.csv` | Each meal service transaction, including meal type and quantity |
+| `laundry_bookings` | `laundry_bookings_rows.csv` | Laundry slot reservations with date, slot label, and status |
+| `shower_reservations` | `shower_reservations_rows.csv` | Shower slot reservations with date, slot label, and status |
+| `haircut_visits` | `haircut_visits_rows.csv` | Individual haircut service visits with timestamps |
+| `bicycle_repairs` | `bicycle_repairs_rows.csv` | Bicycle repair requests with repair types and status |
+| `donations` | `donations_rows.csv` | Donation records by type, item, weight, and donor |
+| `daily_notes` | `daily_notes_rows.csv` | Staff notes by date and service type |
+| `blocked_slots` | `blocked_slots_rows.csv` | Dates and times when service slots were unavailable |
+
+### Schema
+
+**guests**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `external_id` | string | External reference ID |
+| `housing_status` | string | Guest's housing situation |
+| `age_group` | string | Age bracket |
+| `gender` | string | Gender |
+| `location` | string | General location or neighborhood |
+| `notes` | string | Staff notes |
+| `bicycle_description` | string | Description of guest's bicycle, if any |
+| `ban_reason` | string | Reason for any service ban |
+
+**guest_proxies**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `guest_id` | uuid | Guest receiving the service |
+| `proxy_id` | uuid | Guest picking up on their behalf |
+
+**meal_attendance**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `guest_id` | uuid | Guest served |
+| `picked_up_by_guest_id` | uuid | Guest who physically picked up the meal (may differ from guest_id) |
+| `meal_type` | string | Type of meal (e.g., breakfast, lunch) |
+| `quantity` | integer | Number of meals |
+| `served_on` | date | Date of service |
+
+**laundry_bookings**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `guest_id` | uuid | Guest who booked |
+| `scheduled_for` | date | Booking date |
+| `slot_label` | string | Time slot identifier |
+| `status` | string | Booking status (e.g., completed, no-show, cancelled) |
+
+**shower_reservations**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `guest_id` | uuid | Guest who reserved |
+| `scheduled_for` | date | Reservation date |
+| `slot_label` | string | Time slot identifier |
+| `status` | string | Reservation status |
+
+**haircut_visits**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `guest_id` | uuid | Guest served |
+| `served_at` | datetime | Timestamp of service |
+| `service_date` | date | Date of service |
+
+**bicycle_repairs**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `guest_id` | uuid | Guest requesting repair |
+| `requested_at` | datetime | Timestamp of request |
+| `repair_types` | string | Types of repairs requested |
+| `status` | string | Repair status |
+
+**donations**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `donation_type` | string | Category of donation |
+| `item_name` | string | Specific item donated |
+| `weight_lbs` | float | Weight in pounds |
+| `donor` | string | Donor name or identifier |
+| `donated_at` | datetime | Timestamp of donation |
+
+**daily_notes**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `note_date` | date | Date the note was written |
+| `service_type` | string | Service the note refers to |
+| `note_text` | string | Staff note content |
+
+**blocked_slots**
+
+| Field | Type | Description |
+|---|---|---|
+| `id` | uuid | Primary key |
+| `service_type` | string | Service type affected |
+| `slot_time` | string | Time slot that was blocked |
+| `date` | date | Date of the blocked slot |
+
+### Entity Relationship Diagram
 
 ```mermaid
 erDiagram
@@ -149,4 +293,121 @@ erDiagram
     GUESTS ||--o{ SHOWER_RESERVATIONS : "guest_id"
     GUESTS ||--o{ HAIRCUT_VISITS : "guest_id"
     GUESTS ||--o{ BICYCLE_REPAIRS : "guest_id"
-```
+```
+
+## Dataset Creation
+
+### Curation Rationale
+
+Hope's Corner staff collect service data as part of daily operations, but that data has rarely been used for analysis. This dataset was exported and anonymized to support two specific research questions: forecasting how many meals will be needed on a given day, and understanding whether all guest populations have equal access to services. Both have direct operational value for a small non-profit with limited staff and resources.
+
+### Source Data
+
+#### Data Collection and Processing
+
+Records were exported from Hope's Corner's internal management system. Guest identifiers were replaced with random UUIDs prior to export. No names, contact information, or other direct identifiers are present in the dataset. Each table was exported as a separate CSV file.
+
+#### Who are the source data producers?
+
+Data were recorded by Hope's Corner staff during normal service operations in Mountain View, CA.
+
+### Annotations
+
+#### Annotation process
+
+No annotations have been applied. All fields reflect operational records as entered by staff.
+
+#### Who are the annotators?
+
+N/A
+
+#### Personal and Sensitive Information
+
+This dataset covers a vulnerable population. Although guest records are anonymized with UUIDs, the combination of housing status, age group, gender, location, and service history could in principle narrow down individuals in a small community. Handle with care:
+
+- Do not attempt to re-identify guests.
+- Do not share individual-level records outside of research or analysis contexts covered by your data use agreement with Hope's Corner.
+- The `notes` and `ban_reason` fields in `guests` may contain sensitive staff observations and should be treated accordingly.
+
+## Bias, Risks, and Limitations
+
+- **Self-selected population:** This dataset only covers guests who visited Hope's Corner. It does not represent all unhoused or low-income individuals in Mountain View.
+- **Recording gaps:** Not every interaction may have been logged, particularly during busy service periods. Attendance counts may undercount actual demand.
+- **Demographic fields are categorical:** `housing_status`, `age_group`, and `gender` use a fixed set of categories that may not capture the full range of guest situations.
+- **Proxy pickups:** Meal attendance records can reflect a proxy picking up for multiple guests, which requires care when counting unique individuals served.
+- **Seasonal and event variation:** Meal demand is affected by weather, holidays, and local events. Models that don't account for these factors will likely underperform.
+
+### Recommendations
+
+Join `meal_attendance` to `guests` through `guest_id` to segment attendance by demographics. When modeling demand, incorporate external data sources such as weather records and a holiday calendar. Use `blocked_slots` and `daily_notes` to identify and exclude or annotate days with unusual operational circumstances.
+
+## Citation [optional]
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary
+
+| Term | Definition |
+|---|---|
+| **UUID** | Universally Unique Identifier, used here as an anonymized guest key |
+| **Proxy** | A guest who picks up a meal or service on behalf of another registered guest |
+| **Slot label** | The identifier for a specific time block within a service day (e.g., "9:00 AM") |
+| **Housing status** | A categorical field describing a guest's current living situation (e.g., unhoused, transitional, low-income housed) |
+| **Blocked slot** | A time slot that was unavailable for bookings due to staffing, closures, or other operational reasons |
+
+## More Information
+
+### Loading the Dataset
+
+Each table is available as a separate configuration:
+
+```python
+from datasets import load_dataset
+
+# Load guest records
+guests = load_dataset("org/hopescorner", "guests")
+
+# Load meal attendance
+meals = load_dataset("org/hopescorner", "meal_attendance")
+
+# Load laundry bookings
+laundry = load_dataset("org/hopescorner", "laundry_bookings")
+```
+
+### Joining Tables with pandas
+
+```python
+import pandas as pd
+
+guests = pd.read_csv("data/guests_rows.csv")
+meals = pd.read_csv("data/meal_attendance_rows.csv")
+laundry = pd.read_csv("data/laundry_bookings_rows.csv")
+
+# Meal counts per guest with demographics
+meal_counts = (
+    meals.groupby("guest_id")["quantity"]
+    .sum()
+    .reset_index()
+    .rename(columns={"quantity": "total_meals"})
+)
+guest_meals = guests.merge(meal_counts, left_on="id", right_on="guest_id", how="left")
+
+# Daily meal totals (for demand forecasting)
+meals["served_on"] = pd.to_datetime(meals["served_on"])
+daily_totals = meals.groupby("served_on")["quantity"].sum().reset_index()
+daily_totals.columns = ["date", "total_meals"]
+```
+
+## Dataset Card Authors
+
+Hope's Corner, Mountain View, CA
+
+## Dataset Card Contact
+
+[More Information Needed]