The fit method
- Machine Learning course
- Intermediate
- Lesson 3/112
- 5 min
The fit() method learns the model from training data.
fit: the moment of learning
fit(X, y) changes the estimator's state and returns self. Unsupervised estimators take only X. Some models also accept sample_weight for weighted learning:
m = LogisticRegression().fit(X, y)
KMeans(n_clusters=3, n_init=10).fit(X) # no y
m.fit(X, y, sample_weight=agirliklar) # some examples matter morefit validates its inputs: a shape mismatch, NaN or text values produce understandable errors.
fit_transform and n_features_in_
For transformers, fit_transform does fit + transform in one step (often faster). After fit, n_features_in_ stores the expected number of features; calling predict with a different number of columns raises an error:
sc = StandardScaler()
Z = sc.fit_transform(X_train)
print(sc.n_features_in_) # 4
# sc.transform(X_train[:, :3]) ValueError: 4 features expectedfit_transform is only used on training data; transform is always applied to test data.
Practice Weighted training
Train a LinearRegression on the same data first without weights, then giving the last example a weight of 100. Print the two slopes (coef_[0]) with round(..., 3) on separate lines.
Starter code
import numpy as np
from sklearn.linear_model import LinearRegression
X = np.array([[1], [2], [3], [4]])
y = np.array([1, 2, 3, 10])
w = np.array([1, 1, 1, 100])
# buraya yazSample solution
import numpy as np
from sklearn.linear_model import LinearRegression
X = np.array([[1], [2], [3], [4]])
y = np.array([1, 2, 3, 10])
w = np.array([1, 1, 1, 100])
a = LinearRegression().fit(X, y)
b = LinearRegression().fit(X, y, sample_weight=w)
print(round(a.coef_[0], 3))
print(round(b.coef_[0], 3))Expected output
2.8 3.56
Questions and answers
Review the lesson questions together with their answers. To test yourself, solve the lesson interactively.
1 What does a <code>fit</code> call return?
- None
- The estimator itself Correct answer
- The predictions
- The score
Explanation: That is why the chain m = Model().fit(X, y) can be written.
2 Train while weighting the examples.
m.fit(X, y, ____=w)m.fit(X, y, sample_weight=w)Explanation: sample_weight scales each example's contribution to the loss.
3 Match the attributes and methods.
| Term | Meaning |
|---|---|
fit |
Learns, returns self |
fit_transform |
Learn and transform |
n_features_in_ |
Expected column count |
sample_weight |
Example weights |