Statistics

<Model>Statistics allows access to the statistics of an sklearn.<Model>. It is a convenience function that will collect a simple set of statistics.

Consider the same scripts shown in the model section, which fit and predict with the diabetes dataset. You may use the <Model>Statistics to obtain the statistics of the model. This convenience functionality is provided out of the box, but we recommend that you customize how the model is interpreted to extract statistics.

Dummy

Python

import json

from nextmv_sklearn import dummy

# Model code here.

statistics = dummy.DummyRegressorStatistics(fit, X, y, run_duration_start=start_time)
print(json.dumps(statistics.to_dict(), indent=2))

Run the script:

Bash

python main.py
{
  "run": {
    "duration": 0.0009911060333251953
  },
  "result": {
    "custom": {
      "score": 0.0
    }
  },
  "series_data": {},
  "schema": "v1"
}

Ensemble

Python

import json

from nextmv_sklearn import ensemble

# Model code here.

statistics = ensemble.GradientBoostingRegressorStatistics(fit, X, y, run_duration_start=start_time)
print(json.dumps(statistics.to_dict(), indent=2))

Run the script:

Bash

python main.py
{
  "run": {
    "duration": 0.06733989715576172
  },
  "result": {
    "custom": {
      "depth": 3,
      "feature_importances_": [
        0.04838302190078237,
        0.013922652165423995,
        0.261283146271802,
        0.10170455309660177,
        0.028599216988454114,
        0.04562385785680753,
        0.040315277921146336,
        0.01589286353541712,
        0.402315668430028,
        0.04195974183353675
      ],
      "score": 0.7990392018966865
    }
  },
  "series_data": {},
  "schema": "v1"
}

Python

import json

from nextmv_sklearn import ensemble

# Model code here.

statistics = ensemble.RandomForestRegressorStatistics(fit, X, y, run_duration_start=start_time)
print(json.dumps(statistics.to_dict(), indent=2))

Run the script:

Bash

python main.py
{
  "run": {
    "duration": 0.15362906455993652
  },
  "result": {
    "custom": {
      "feature_importances_": [
        0.05804547555574568,
        0.011942445816542004,
        0.26046692216365497,
        0.09831628970586574,
        0.04414312280601346,
        0.05823550762545449,
        0.05163041383919266,
        0.024326122944217127,
        0.323262101101696,
        0.06963159844161786
      ],
      "score": 0.9180285142096314
    }
  },
  "series_data": {},
  "schema": "v1"
}

Linear model

Python

import json

from nextmv_sklearn import linear_model

# Model code here.

statistics = linear_model.LinearRegressionStatistics(fit, X, y, run_duration_start=start_time)
print(json.dumps(statistics.to_dict(), indent=2))

Run the script:

Bash

python main.py
{
  "run": {
    "duration": 0.0011751651763916016
  },
  "result": {
    "custom": {
      "score": 0.5177484222203499
    }
  },
  "series_data": {},
  "schema": "v1"
}

Neural network

Python

import json

from nextmv_sklearn import neural_network

# Model code here.

statistics = neural_network.MLPRegressorStatistics(fit, X, y, run_duration_start=start_time)
print(json.dumps(statistics.to_dict(), indent=2))

Run the script:

Bash

$ python main.py -max_iter 2500
{
  "run": {
    "duration": 0.9213109016418457
  },
  "result": {
    "custom": {
      "score": 0.5105739729721858
    }
  },
  "series_data": {},
  "schema": "v1"
}

Tree

Python

import json

from nextmv_sklearn import tree

# Model code here.

statistics = tree.DecisionTreeRegressorStatistics(fit, X, y, run_duration_start=start_time)
print(json.dumps(statistics.to_dict(), indent=2))

Run the script:

Bash

$ python main.py
{
  "run": {
    "duration": 0.0031609535217285156
  },
  "result": {
    "custom": {
      "depth": 20,
      "feature_importances_": [
        0.03648268541156411,
        0.008967396431459116,
        0.23191566128391577,
        0.082693316388564,
        0.07347510143597435,
        0.056688846730468985,
        0.07264156168631583,
        0.013709115567881148,
        0.3487799461424682,
        0.07464636892138855
      ],
      "n_leaves": 433,
      "score": 1.0
    }
  },
  "series_data": {},
  "schema": "v1"
}

Learn how to work with Statistics.

Dummy

Ensemble

Linear model

Neural network

Tree

Model

Solution

Contents