Uniq

The uniq function is used to estimate the count of unique values in a dataset. It's particularly useful for analyzing large datasets where an exact count may not be necessary or where performance is a concern. Below is an example query using the uniq function within the context of EVM TokenHolders API to get the number of unique token holders for a specific token on a given date.

Example Query

{
  EVM(dataset: archive, network: eth) {
    Holders(
      date: "2024-03-04"
      where: { Currency: { SmartContract: { is: "0x95AD61B0A150D79219DCF64E1E6CC01F0B64C4CE" } }, Balance: {Amount: {gt: "0"}}}
    ) {
      exact: uniq(of: Holder_Address, method: exact)
      count(distinct: Holder_Address)
      approximate: uniq(of: Holder_Address, method: approximate)
    }
  }
}

Result

{
  "EVM": {
    "TokenHolders": [
      {
        "approximate": "1373089",
        "count": "1378364",
        "exact": "1378364"
      }
    ]
  }
}

Understanding Uniq Function Variants

• Estimate (uniq): The uniq function estimates the count of unique values. It uses an adaptive sampling algorithm. This approach is highly accurate and CPU-efficient for processing large datasets.
• Exact Count (uniqExact): For scenarios requiring precision, the uniqExact function calculates the exact number of unique values. Although more memory-intensive it guarantees accuracy.

When to Use Each Variant

• Use uniq for Estimates: Opt for the uniq function when an approximate count suffices. This function is ideal for large datasets where performance and efficiency are priorities.
• Use uniqExact for Precision: Choose the uniqExact function when accuracy is non-negotiable. Keep in mind the potential for increased memory usage.