joaopcnogueira/mletrics

Machine learning metrics that are not easy to found

mletrics

from mletrics.stability import psi
from mletrics.classification import ks

Install

pip install mletrics

How to use

Calculating psi values

import numpy as np
import pandas as pd
from sklearn.pipeline import Pipeline
from sklearn.impute import SimpleImputer
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import OneHotEncoder
from scikitplot.metrics import plot_ks_statistic
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split

from pathlib import Path

p = Path('..')
df = pd.read_csv(p/'datasets/titanic.csv')
df.head()

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	PassengerId	Survived	Pclass	Name	Sex	Age	SibSp	Parch	Ticket	Fare	Cabin	Embarked
0	1	0	3	Braund, Mr. Owen Harris	male	22.0	1	0	A/5 21171	7.2500	NaN	S
1	2	1	1	Cumings, Mrs. John Bradley (Florence Briggs Th...	female	38.0	1	0	PC 17599	71.2833	C85	C
2	3	1	3	Heikkinen, Miss. Laina	female	26.0	0	0	STON/O2. 3101282	7.9250	NaN	S
3	4	1	1	Futrelle, Mrs. Jacques Heath (Lily May Peel)	female	35.0	1	0	113803	53.1000	C123	S
4	5	0	3	Allen, Mr. William Henry	male	35.0	0	0	373450	8.0500	NaN	S

cat_vars = ['Pclass', 'Sex', 'Embarked']
num_vars = ['Age', 'SibSp', 'Fare']
features = cat_vars + num_vars
target = 'Survived'

X = df[features].copy()
y = df[target].copy()

num_pipe = Pipeline(steps=[
        ('imputer', SimpleImputer(strategy='constant', fill_value=-999))
])

cat_pipe = Pipeline(steps=[
        ('imputer', SimpleImputer(strategy='constant', fill_value='missing')),
        ('ohe', OneHotEncoder(sparse=False, handle_unknown='ignore'))
]) 

transformers = ColumnTransformer(transformers=[
                ('numeric', num_pipe, num_vars),
                ('categoric', cat_pipe, cat_vars)
])

model = Pipeline(steps=[
        ('transformers', transformers),
        ('model', RandomForestClassifier(random_state=42, max_depth=3))
])

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

model.fit(X_train, y_train)

y_proba_train = model.predict_proba(X_train)[:,1]
y_proba_test  = model.predict_proba(X_test)[:,1]

calculate psi value for the model probability between train and test

psi(y_proba_train, y_proba_test)

0.06001324825109782

• PSI < 0.1 - No change. You can continue using existing model.
• PSI >= 0.1 but less than 0.2 - Slight change is required.
• PSI >= 0.2 - Significant change is required. Ideally, you should not use this model any more.

Reference: https://www.listendata.com/2015/05/population-stability-index.html

Calculating the KS

ks(y_test, y_proba_test)

0.5886743886743887

Comparing with the ks calculated by scikit-plot:

plot_ks_statistic(y_test, np.column_stack([1-y_proba_test, y_proba_test]))

<AxesSubplot:title={'center':'KS Statistic Plot'}, xlabel='Threshold', ylabel='Percentage below threshold'>