ロジスティック回帰¶

ロジスティック回帰全般については以下を参照してください。 https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression

データとモジュールのロード

import pandas as pd
from sklearn import model_selection

data = pd.read_csv("input/pn_same_judge_preprocessed.csv")
train, test = model_selection.train_test_split(data, test_size=0.1, random_state=0)

from sklearn.pipeline import Pipeline
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics import ConfusionMatrixDisplay
from sklearn.metrics import PrecisionRecallDisplay

LogisticRegression¶

sklearn.linear_model.LogisticRegression を使います。

from sklearn.linear_model import LogisticRegression

pipe = Pipeline([
    ("vect", TfidfVectorizer(tokenizer=str.split)),
    ("clf", LogisticRegression())
])

pipe.fit(train["tokens"], train["label_num"])

Pipeline(steps=[('vect',
                 TfidfVectorizer(tokenizer=<method 'split' of 'str' objects>)),
                ('clf', LogisticRegression())])

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

pred = pipe.predict(test["tokens"])
ConfusionMatrixDisplay.from_predictions(y_true=test["label_num"], y_pred=pred)

<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7f6a74c22d60>

score = pipe.predict_proba(test["tokens"])[:,1]

PrecisionRecallDisplay.from_predictions(
    y_true=test["label_num"],
    y_pred=score,
    name="LogisticRegression",
)

<sklearn.metrics._plot.precision_recall_curve.PrecisionRecallDisplay at 0x7f6a74b668b0>

不均衡データに対応する¶

class_weight パラメータで不均衡データに対応できます。

pipe_weight = Pipeline([
    ("vect", TfidfVectorizer(tokenizer=str.split)),
    ("clf", LogisticRegression(class_weight="balanced"))
])

pipe_weight.fit(train["tokens"], train["label_num"])

Pipeline(steps=[('vect',
                 TfidfVectorizer(tokenizer=<method 'split' of 'str' objects>)),
                ('clf', LogisticRegression(class_weight='balanced'))])

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

score_weight = pipe_weight.predict_proba(test["tokens"])[:,1]

class_weightオプションを付けないモデルと比較します。

import  matplotlib.pyplot as plt

_, ax = plt.subplots()
for name, pred in [
    ("LogisticRegression", score),
    ("LogisticRegression+weight", score_weight),
]:
    PrecisionRecallDisplay.from_predictions(ax=ax, y_true=test["label_num"], y_pred=pred, name=name)

NLP クックブック

ロジスティック回帰

Contents

ロジスティック回帰¶

LogisticRegression¶

不均衡データに対応する¶