README.md

---
language: en
license: cc-by-4.0
tags:
- business
- economic
- yelp
- tabular
- regression
- classification
dataset_info:
  features:
  - name: business_id
    dtype: string
  - name: rating_x_reviews
    dtype: float64
  - name: review_count
    dtype: int64
  - name: num_categories
    dtype: int64
  - name: years_in_business
    dtype: float64
  - name: num_checkins
    dtype: int64
  - name: has_checkin
    dtype: int64
  - name: is_open
    dtype: int64
  - name: latitude
    dtype: float64
  - name: longitude
    dtype: float64
  - name: fips
    dtype: string
  - name: pcpi
    dtype: float64
  - name: poverty_rate
    dtype: float64
  - name: median_household_income
    dtype: float64
  - name: unemployment_rate
    dtype: float64
  - name: avg_weekly_wages
    dtype: float64
  - name: cat_*
    dtype: int64
  splits:
  - name: train
    num_examples: 17000+
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

# Yelp Business Economic Indicators Dataset

## ⚠️ Important Version Notice (Please Read)

**Version 3 (v3) should be used instead of v1 or v2.**

- **v1 (~1k rows)** and **v2 (~10k rows)** use county-level economic indicators from **2023**
- **v3 (~17k rows)** uses economic indicators from **2018**, which better aligns with:
  - the temporal coverage of the Yelp Open Dataset
  - business survival modeling
  - avoidance of temporal leakage

Earlier versions are retained for reproducibility only and are **not recommended** for modeling.

---

## Dataset Overview

This dataset combines business-level information from the **Yelp Open Dataset** with **county-level economic indicators** sourced from U.S. government datasets.

The dataset is designed for **predictive modeling**, particularly tasks such as:

- Predicting whether a business will remain open or close
- Studying business survival and risk
- Analyzing interactions between local economic conditions and business outcomes

Each row corresponds to a single Yelp business.

---

## What’s New in v3

Version 3 adds substantial feature improvements over v1 and v2:

- ✔ Expanded dataset size (~17k rows)
- ✔ Corrected economic data timing (2018 instead of 2023)
- ✔ Yelp category multi-hot encoded features
- ✔ Business longevity features derived from Yelp check-ins
- ✔ Improved engagement signals

---

## Features (v3)

### Business Engagement & Quality
| Column | Description |
|------|------------|
| `rating_x_reviews` | Yelp rating multiplied by log(review_count + 1) |
| `review_count` | Total number of Yelp reviews |
| `num_categories` | Number of Yelp categories assigned |

---

### Business Longevity (Derived from Check-ins)
| Column | Description |
|------|------------|
| `years_in_business` | Years between first and last Yelp check-in (observed lifespan) |
| `num_checkins` | Total number of Yelp check-ins |
| `has_checkin` | 1 if any check-in exists, else 0 |

**Note:**  
These features estimate *observed Yelp activity duration*, not true opening date. They are intended for **relative comparison across businesses**, not causal inference.

---

### Target Variable
| Column | Description |
|------|------------|
| `is_open` | 1 if business is open, 0 if closed |

---

### Location
| Column | Description |
|------|------------|
| `latitude` | Business latitude |
| `longitude` | Business longitude |
| `fips` | County FIPS code |

---

### County-Level Economic Indicators (2018)
| Column | Description |
|------|------------|
| `pcpi` | Per capita personal income (USD) |
| `poverty_rate` | Poverty rate (%) |
| `median_household_income` | Median household income (USD) |
| `unemployment_rate` | Unemployment rate (%) |
| `avg_weekly_wages` | Average weekly wages (USD) |

---

### Yelp Category Indicators
Multi-hot encoded binary features indicating whether a business belongs to a given category.

Examples:
- `cat_Restaurants`
- `cat_Food`
- `cat_Automotive`
- `cat_Bars`
- `cat_Health & Medical`
- `cat_Shopping`
- …

Only the **most frequent categories** are included to limit sparsity.

---

## Normalization Note

Some earlier versions of this dataset were normalized for convenience.  
**Normalization is not required and not recommended for tree-based models** such as:

- XGBoost
- LightGBM

Version 3 is intended to be used **as-is with raw feature values**.

---

## Sources

### Yelp
- **Yelp Open Dataset**  
  https://www.yelp.com/dataset  
  Business attributes, categories, reviews, and check-ins.

### Economic Indicators
- **BLS – Quarterly Census of Employment and Wages (QCEW)**  
  https://www.bls.gov/cew/  
  Average weekly wages.

- **FRED – Federal Reserve Economic Data**  
  https://fred.stlouisfed.org/  
  Per capita personal income (PCPI).

- **U.S. Census Bureau – ACS / SAIPE**  
  https://www.census.gov/programs-surveys/saipe.html  
  Poverty rates, median household income, unemployment rates.

---

## Usage Notes

- Missing values may exist for some businesses (e.g., no check-ins)
- No feature normalization is required for tree-based models
- Designed for tabular ML, feature interaction modeling, and exploratory analysis

---

## Example Usage

```python
import pandas as pd
from datasets import load_dataset

# Load from Hugging Face
dataset = load_dataset("your-username/yelp-business-economic-indicators")
df = dataset["train"].to_pandas()

print(df.head())

```

## License

This dataset is released under the CC BY 4.0 License.

Yelp Dataset Terms: https://www.yelp.com/dataset/terms

Economic data sources are U.S. government public-domain datasets.