Click to Flip Card
Logistic Regression
Is Logistic Regression used for linear predictions? What does Logistic Regression predict? How do we compare different values with varying scales?
Classification
Despite the name, logistic regression is actually a widely used for classification technique.
""" Logistic Regression (Examp Scores)
Logistic Regression transforms the linear combination of the input features
into a range of probabilites.
Linear Regression predicts continuous numeric values, while
Logistic Regression predicts probabilities for categorical outcomes.
If we can't rely on similarity to make predictions, as with KNN Classifer,
Logistic Regression is a better choice.
Suppose we have two features: exam1 score (continuous) and exam2 score (continuous).
The target variable is student admission (binary: 0/1 not admited or admited).
"""
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
# Dataset
X = [[80, 85], [90, 95], [70, 75], [60, 65]]
y = [1, 1, 0, 0]
# Train and test data
X1, X2, y1, y2 = train_test_split(X, y, test_size=0.2, random_state=42)
# Fitting the model
model = LogisticRegression()
model.fit(X1, y1)
score = model.score(X2, y2)
# Prediction
x_new = [92, 64]
y_pred = model.predict([x_new])
assert y_pred == 1
print("Unknown:", x_new)
print("Prediction:", y_pred)
print("Score:", round(score,2))
"""
Unknown: [92, 64]
Prediction: [1]
Score: 1.0
"""
Scaling
If data have different values, and even different measurement units, we use scalling.
""" Breast cancer / Logistic Regression
As our data have different values, and even different measurement units,
we use scalling in order to compare them.
"""
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.datasets import load_breast_cancer
from sklearn.preprocessing import StandardScaler
# Dataset
dataset = load_breast_cancer()
X = dataset.data
y = dataset.target
# Scalling
X = StandardScaler().fit_transform(X) # Look Here
# Training and test data
X1, X2, y1, y2 = train_test_split(X, y, test_size=0.2, random_state=42)
# Fitting the model
model = LogisticRegression()
model.fit(X1, y1)
score = model.score(X2, y2)
# Predict unknown
X_new = X2[15]
y_pred = model.predict(X_new.reshape(1, -1))
y_expected = dataset['target_names'][y2[15]]
assert y_pred == list(dataset['target_names']).index(y_expected)
# Output
print("Targets:", dataset['target_names'])
print("X_new:\n", X_new)
print("Expected:", y_expected)
print("y_pred:", dataset['target_names'][y_pred][0])
"""
Targets: ['malignant' 'benign']
X_new:
[ 1.41231974 1.62902878 1.52943195 1.35695235 1.7890792 1.41679395
1.31702506 2.52731642 -0.64847556 1.33855706 0.41082896 3.07195015
1.45288552 0.58883002 0.16801384 1.95735569 -0.34992981 1.07607669
1.21292827 2.49460396 0.84092285 1.14687248 1.01387382 0.73378242
0.29946077 0.17452465 -0.13907293 1.05815376 -0.95409536 0.4479916 ]
Expected: malignant
y_pred: malignant
"""
Last update: 208 days ago
