HROCH


NameHROCH JSON
Version 1.0.11 PyPI version JSON
download
home_pagehttps://github.com/janoPig/HROCH/
SummarySymbolic regression
upload_time2022-09-25 09:32:26
maintainer
docs_urlNone
authorJano
requires_python
licenseMIT
keywords machine-learning numpy symbolic-regression fuzzy
VCS
bugtrack_url
requirements No requirements were recorded.
Travis-CI No Travis.
coveralls test coverage No coveralls.
            
# HROCH  

**High optimized symbolic regression.**

- Zero hyperparameter tunning.
- Accurate results in seconds or minutes, in contrast to slow GP-based methods.
- Small models size.
- Good results with noise data.
- Support mathematic equations and fuzzy logic operators.
- Support 32 and 64 bit floating point arithmetic.
- Work with unprotected version of math operators (log, sqrt, division)
- CLI

## Dependencies

- AVX2 instructions set(all modern CPU support this)
- numpy

## Installation

```sh
pip install HROCH
```

## Usage

```python
from HROCH import PHCRegressor

reg = PHCRegressor(numThreads=8, timeLimit=60.0, problem='math', precision='f64')
reg.fit(X_train, y_train)
yp = reg.predict(X_test)
# print symbolic expression
print(reg.sexpr)
```

## Performance

### Feynman dataset

Approximate comparison with methods tested in [srbench](https://cavalab.org/srbench/results/#results-for-ground-truth-problems).

|Algorithm|Training time (s)|Model size|R2 > 0.999|R2 > 0.999999|R2 > 0.999999999|R2 mean          |
|---------|----------------:|---------:|:--------:|:-----------:|:--------------:|:---------------:|
|MRGP     |14893            |3177      |0.931     |0.000        |0.000           |0.998853549755939|
|Operon   |2093             |70        |0.862     |0.655        |0.392           |0.990832974928022|
|AIFeynman|26822            |121       |0.785     |0.689        |0.680           |0.923670858619585|
|__HROCH__|__42__               |__17__        |__0.781__     |__0.679__        |__0.633__           |__0.988862822072670__|
|SBP-GP   |28944            |487       |0.737     |0.388        |0.246           |0.994645420032544|
|GP-GOMEA |3677             |34        |0.716     |0.539        |0.504           |0.996850949284431|
|AFP_FE   |17682            |41        |0.591     |0.315        |0.185           |0.985876419645066|
|EPLEX    |10599            |56        |0.470     |0.121        |0.082           |0.991763792716299|
|AFP      |2895             |37        |0.448     |0.263        |0.159           |0.968488776363814|
|FEAT     |1561             |195       |0.397     |0.121        |0.112           |0.932465581448533|
|gplearn  |3716             |78        |0.328     |0.151        |0.151           |0.901020570640627|
|ITEA     |1435             |21        |0.276     |0.233        |0.224           |0.911713461958873|
|DSR      |615              |15        |0.250     |0.207        |0.207           |0.875784840006460|
|BSR      |28800            |25        |0.108     |0.073        |0.043           |0.693995349495648|
|FFX      |19               |268       |0.000     |0.000        |0.000           |0.908164756903951|


            

Raw data

            {
    "_id": null,
    "home_page": "https://github.com/janoPig/HROCH/",
    "name": "HROCH",
    "maintainer": "",
    "docs_url": null,
    "requires_python": "",
    "maintainer_email": "",
    "keywords": "machine-learning,numpy,symbolic-regression,fuzzy",
    "author": "Jano",
    "author_email": "hroch.regression@gmail.com",
    "download_url": "https://files.pythonhosted.org/packages/82/d0/4fbe531488fbe3f9ddb6df3eeb1c3c43cdb7355c6ffeb5b521bc1d3bd3dd/HROCH-1.0.11.tar.gz",
    "platform": null,
    "description": "\n# HROCH  \n\n**High optimized symbolic regression.**\n\n- Zero hyperparameter tunning.\n- Accurate results in seconds or minutes, in contrast to slow GP-based methods.\n- Small models size.\n- Good results with noise data.\n- Support mathematic equations and fuzzy logic operators.\n- Support 32 and 64 bit floating point arithmetic.\n- Work with unprotected version of math operators (log, sqrt, division)\n- CLI\n\n## Dependencies\n\n- AVX2 instructions set(all modern CPU support this)\n- numpy\n\n## Installation\n\n```sh\npip install HROCH\n```\n\n## Usage\n\n```python\nfrom HROCH import PHCRegressor\n\nreg = PHCRegressor(numThreads=8, timeLimit=60.0, problem='math', precision='f64')\nreg.fit(X_train, y_train)\nyp = reg.predict(X_test)\n# print symbolic expression\nprint(reg.sexpr)\n```\n\n## Performance\n\n### Feynman dataset\n\nApproximate comparison with methods tested in [srbench](https://cavalab.org/srbench/results/#results-for-ground-truth-problems).\n\n|Algorithm|Training time (s)|Model size|R2 > 0.999|R2 > 0.999999|R2 > 0.999999999|R2 mean          |\n|---------|----------------:|---------:|:--------:|:-----------:|:--------------:|:---------------:|\n|MRGP     |14893            |3177      |0.931     |0.000        |0.000           |0.998853549755939|\n|Operon   |2093             |70        |0.862     |0.655        |0.392           |0.990832974928022|\n|AIFeynman|26822            |121       |0.785     |0.689        |0.680           |0.923670858619585|\n|__HROCH__|__42__               |__17__        |__0.781__     |__0.679__        |__0.633__           |__0.988862822072670__|\n|SBP-GP   |28944            |487       |0.737     |0.388        |0.246           |0.994645420032544|\n|GP-GOMEA |3677             |34        |0.716     |0.539        |0.504           |0.996850949284431|\n|AFP_FE   |17682            |41        |0.591     |0.315        |0.185           |0.985876419645066|\n|EPLEX    |10599            |56        |0.470     |0.121        |0.082           |0.991763792716299|\n|AFP      |2895             |37        |0.448     |0.263        |0.159           |0.968488776363814|\n|FEAT     |1561             |195       |0.397     |0.121        |0.112           |0.932465581448533|\n|gplearn  |3716             |78        |0.328     |0.151        |0.151           |0.901020570640627|\n|ITEA     |1435             |21        |0.276     |0.233        |0.224           |0.911713461958873|\n|DSR      |615              |15        |0.250     |0.207        |0.207           |0.875784840006460|\n|BSR      |28800            |25        |0.108     |0.073        |0.043           |0.693995349495648|\n|FFX      |19               |268       |0.000     |0.000        |0.000           |0.908164756903951|\n\n",
    "bugtrack_url": null,
    "license": "MIT",
    "summary": "Symbolic regression",
    "version": "1.0.11",
    "split_keywords": [
        "machine-learning",
        "numpy",
        "symbolic-regression",
        "fuzzy"
    ],
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "c433d009fbc588aaeea305540e5fa993",
                "sha256": "f9c33466ae3b5fa39931c48c075f561dd4b433650fe14750d361e4375d7e10dc"
            },
            "downloads": -1,
            "filename": "HROCH-1.0.11-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "c433d009fbc588aaeea305540e5fa993",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 1240183,
            "upload_time": "2022-09-25T09:32:25",
            "upload_time_iso_8601": "2022-09-25T09:32:25.025341Z",
            "url": "https://files.pythonhosted.org/packages/77/10/4a8a6e7d0f5bad2f9abe95ffbff45ca409e35a7860540274f846eac0fa0e/HROCH-1.0.11-py3-none-any.whl",
            "yanked": false,
            "yanked_reason": null
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "69f52902a6dd16d50357e9555de7795c",
                "sha256": "c6c7ca9f4699310f714586bcb3161c13ef6b359f1203122472003da8403aba84"
            },
            "downloads": -1,
            "filename": "HROCH-1.0.11.tar.gz",
            "has_sig": false,
            "md5_digest": "69f52902a6dd16d50357e9555de7795c",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 1237993,
            "upload_time": "2022-09-25T09:32:26",
            "upload_time_iso_8601": "2022-09-25T09:32:26.989090Z",
            "url": "https://files.pythonhosted.org/packages/82/d0/4fbe531488fbe3f9ddb6df3eeb1c3c43cdb7355c6ffeb5b521bc1d3bd3dd/HROCH-1.0.11.tar.gz",
            "yanked": false,
            "yanked_reason": null
        }
    ],
    "upload_time": "2022-09-25 09:32:26",
    "github": true,
    "gitlab": false,
    "bitbucket": false,
    "github_user": "janoPig",
    "github_project": "HROCH",
    "travis_ci": false,
    "coveralls": false,
    "github_actions": true,
    "lcname": "hroch"
}
        
Elapsed time: 0.53767s