# Dataset Join Loss Preview

> Dataset Join Loss Preview is a paid API for AI agents from www.mahastrategies.com, paid per call via x402, $0.035019/call, status unknown (last checked 2026-10-19).

Measures join multiplication factor and unmatched record counts when joining two datasets on declared key fields, without returning the joined data itself.

## Facts

- Endpoint: POST https://www.mahastrategies.com/api/v1/compute/datasets/join-loss?utm_source=zero.xyz
- Price: $0.035019/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-19
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/dataset-join-loss-preview-458ab2f6
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_5hYA_RUV5IKjrClP2ExAz

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability dataset-join-loss-preview-458ab2f6 -d '<json body>'
```

Example prompt: Before I merge these two datasets on 'user_id' and 'region', can you preview the join loss — how many records won't match and whether there's any row multiplication — using synthetic data class? Left dataset has up to 200 records and right has up to 150.

## When to prefer this

Use this endpoint when you need to assess join quality and data loss risk before committing to a full join operation, especially when datasets are large or downstream pipelines are costly. Prefer this over a full join endpoint when you only need diagnostics (match rate, fan-out factor, unmatched counts) rather than the actual joined rows. Ideal for ETL pre-flight checks, data quality audits, and key-field selection decisions. Restricted to declared key fields only — not suitable for fuzzy or probabilistic matching.

## Known failure modes

- Dataset arrays exceed 500-item limit — request rejected with validation error
- keyFields array empty or exceeds 8 fields — rejected with schema validation error
- dataClass value not 'synthetic' or 'public' — enum validation error
- Key fields referenced in keyFields not present in dataset records — mismatch error or empty match result
- Malformed record objects in left or right arrays — parsing error

## How this service works

Measure join multiplication and unmatched records without returning joined cells. Declared keys only.

## Output

Returns join quality metrics including the multiplication factor (how many times rows fan out), count of unmatched records from the left dataset, count of unmatched records from the right dataset, and key match statistics — without returning any of the actual joined cell data.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "left": {
   "type": "array",
   "maxItems": 500
  },
  "right": {
   "type": "array",
   "maxItems": 500
  },
  "dataClass": {
   "enum": [
    "synthetic",
    "public"
   ],
   "type": "string"
  },
  "keyFields": {
   "type": "array",
   "maxItems": 8,
   "minItems": 1,
   "uniqueItems": true
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/dataset-join-loss-preview-458ab2f6/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from www.mahastrategies.com](https://www.zero.xyz/host/www.mahastrategies.com/llms.txt)