Benchmark: Array duplicate removal for duplicates exceeding `N`-number

Tests:

original duplicate removal
// Remove duplicates that occur 3 or more times in an array // keeping unique values and those with less than 3 function removeMany(arr) { const newArr = Array.from(arr).sort() let count = 0; let result = [] newArr.forEach((value, index, ar) => { count += 1; // refactored afterwards from (ar[index + 1] !== value) if (ar.lastIndexOf(value) <= index && count <= 2) { for (var i = 0; i < arr.length; i++) { if (arr[i] === value) { result.push(arr[i]) } } count = 0 } else if (ar[index + 1] !== value) { count = 0; } }); // +1 is there anyway to return a result that mimicks the original order of `numbers`? return result; // [1, 2, 2, 3, 4, 4] } const numbers = [1, 2, 3, 2, 4, 4, 5, 5, 5, 5]; console.log(removeMany(numbers));
2-pass approach
// Remove duplicates that occur 3 or more times in an array // keeping unique values and those with less than 3 function removeMany(arr) { let countMappings = arr.reduce(function(carry, item) { if (carry[item]!== undefined) { carry[item]++; } else { carry[item] = 1; } return carry; }, {}); return arr.reduce(function(final, item) { if (countMappings[item] <3) { final.push(item); } return final; }, []); } const numbers = [1, 2, 3, 2, 4, 4, 5, 5, 5, 5]; console.log(removeMany(numbers));
2-pass with filtering
function removeMany(numbers, max) { const numberMap = numbers.reduce((map, num) => { map[num] = map[num] ? map[num] + 1 : 1; return map; }, []); return numbers.filter(num => numberMap[num] < max); } const numbers = [1, 2, 3, 2, 4, 4, 5, 5, 5, 5]; console.log(removeMany(numbers, 3));
2pf-obj
function removeMany(numbers, max) { const numberMap = numbers.reduce((map, num) => { if (!map[num]) map[num] = 0; map[num]++; return map; }, {}); return numbers.filter(num => numberMap[num] < max); } const numbers = [1, 2, 3, 2, 4, 4, 5, 5, 5, 5]; console.log(removeMany(numbers, 3));

Rendered benchmark preparation results:

Suite status: <idle, ready to run>

Previous results

Test case name	Result
original duplicate removal
2-pass approach
2-pass with filtering
2pf-obj

Fastest: N/A

Slowest: N/A

Autogenerated LLM Summary (model llama3.2:3b, generated 26 days ago):

LLMs can make mistakes. Check important info.

Let's dive into explaining what is tested in the provided JSON and exploring the different approaches.

**Benchmark Purpose**

The benchmark measures the performance of removing duplicates from an array, where duplicates are considered as values that occur 3 or more times. The goal is to find unique values and those with less than 3 occurrences.

**Approaches Compared**

There are four approaches compared in this benchmark:

1. **Original Duplicate Removal**: This approach uses a simple iterative method to remove duplicates. It keeps the original order of elements.
2. **2-Pass Approach**: This approach involves two passes over the array: one for counting occurrences and another for filtering out duplicates.
3. **2-Pass with Filtering**: Similar to the 2-pass approach, but it filters out duplicates in a single pass using an object mapping.
4. **2PF-Obj** (likely a typo): This approach is similar to the 2-pass with filtering method.

**Pros and Cons of Each Approach**

1. **Original Duplicate Removal**:
	* Pros: Simple, easy to understand, and potentially faster since it doesn't require additional data structures.
	* Cons: May have higher overhead due to the use of `lastIndexOf` for counting occurrences.
2. **2-Pass Approach**:
	* Pros: Effective for large datasets, allows for early filtering, and can be optimized with caching.
	* Cons: Requires two passes over the array, which can increase memory usage and overhead.
3. **2-Pass with Filtering**:
	* Pros: Fast, efficient, and scalable, as it uses an object mapping to filter out duplicates in a single pass.
	* Cons: May have higher memory usage due to the creation of the object map.
4. **2PF-Obj**:
	* Pros: Similar to 2-pass with filtering, but might have slight performance improvements due to the use of an object mapping.
	* Cons: Same as the 2-pass approach (two passes), which can increase memory usage and overhead.

**Library Usage**

None of the approaches explicitly use a library for removing duplicates. However, the `lastIndexOf` method used in the original duplicate removal approach is part of the JavaScript standard library.

**Special JS Features or Syntax**

The benchmark uses an array reduction technique (`reduce`) to count occurrences and filter out duplicates. This is a feature introduced in ECMAScript 2015 (ES6) for functional programming. The use of `const` declarations, arrow functions, and template literals are also part of ES6 syntax.

**Benchmark Results**

The latest benchmark results show the performance differences between these approaches on a Chrome 61 browser:

* **2-Pass with Filtering**: The fastest approach, with an average of 43858 executions per second.
* **2PF-Obj**: Second-fastest, with an average of 41699 executions per second.
* **Original Duplicate Removal**: Third-fastest, with an average of 39272 executions per second.
* **2-Pass Approach**: The slowest approach, with an average of 48679 executions per second.

Overall, the choice of approach depends on the specific requirements and constraints of the use case.

Latest run results:

Run details: (Test run date: 7 years ago)

User agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/61.0.3163.79 Safari/537.36

Browser/OS: Chrome 61 on Windows

View result in a separate tab

Test name	Executions per second
original duplicate removal	39272.0 Ops/sec
2-pass approach	41699.6 Ops/sec
2-pass with filtering	48679.2 Ops/sec
2pf-obj	43858.6 Ops/sec

Related benchmarks:

Array duplicate removal for duplicates exceeding `N`-number (version: 0)

https://codereview.stackexchange.com/q/175067/120114

Comparing performance of: original duplicate removal vs 2-pass approach vs 2-pass with filtering vs 2pf-obj

Created: 7 years ago by: Guest

Jump to the latest result

original duplicate removal

2-pass approach

2-pass with filtering

2pf-obj

Suite status: <idle, ready to run>

Experimental features:

Fastest: N/A

Slowest: N/A

Autogenerated LLM Summary (model llama3.2:3b, generated 26 days ago):