Course1-Week3-分类问题

(i))−y(i)).

本节Quiz：

Wh i ch of t h e follo win g can address ov e rfi t t in g?
√ Colle ct more training data.
× Remo v e a random set of training examples.
√ Ap ply reg ul ar iza t i on.
√ Sel e ct a su b set of t h e more rele v ant features.

You fit logistic reg ression with polyno mial features to a dataset, and your model looks like this. What would you co nclude? (Pic k one)
× Th e model has hig h bias (unde rfit). Thus, adding data is likely to h elp.
√ The model has high variance (overfit). Thus, adding data is likely to help.
× The model has high bias (unde rfit). Thus, adding data is, by itself, unlikely to help much.
× The model has high variance (overfit). Thus, adding data is, by itself, unlikely to help much.

Sup pose you have a regularized linear reg ression model. If you inc rease the regularization parameter

λ

lambda

$λ$ , what do you expect to happen to the parameters

w

1

,

w

2

,

.

.

.

,

w

n

w_1, w_2, …, w_n

$w_{1}, w_{2}, \dots, w_{n}$ ?
√ This will reduce the size of the parameters

w

1

,

w

2

,

.

.

.

,

w

n

w_1, w_2, …, w_n

$w_{1}, w_{2}, \dots, w_{n}$ .
× This will inc rease the size of the parameters

w

1

,

w

2

,

.

.

.

,

w

n

w_1, w_2, …, w_n

$w_{1}, w_{2}, \dots, w_{n}$ .

注：C1_W3_Logistic_Regression包括了逻辑回归、正则化逻辑回归的练习题。