New answers tagged

7

There is some important information missing in your question, i.e. what the standard parameters are and what kind of logistic regression you use. When you use sklearn.linear_model.LogisticRegression, you will see in the docs that the first hyperparameter is the penalty which defaults to l2. This means that by default "shrinkage" of parameters is ...


0

Regularization tries to reduce the chance of overfitting by reducing the sensitivity to small changes in the input data. This is not as much of an issue for the intercept term (relative to the coefficients) so it is often not included.


1

Instead of using the estimator attribute you should be using the best_estimator attribute, after which you can access the underlying estimators of the MultiOutputRegressor using the estimators_ attribute. You can then access the coefficients as follows: coefficients = [estimator.coef_ for estimator in best_model.best_estimator_.estimators_] # [array([-0. ...


0

Allright, here's what I think: I would always recommend splitting the training set if the amount of data allows it. Your train/test split in the train data wil then be a train/validation split. The validation set will allow you to test your model for things such as overfitting on the training data. Even though you can get the ROC/AUC score for the test set ...


0

I’m going to solve a different problem so that you still have to do something on your own for your homework, but this should give you a template for solving your assignment. $$ (X_1,Y_1)=(1,1)\\ (X_2,Y_2)=(3,5) $$ If $h_{\theta}(x)=\theta_0+\theta_1x+\theta_2x^2$, let’s calculate the cost $J(\theta_0,\theta_1,\theta_2)=\sum_i \big(h(x^{(i)}-y^{(i)} \big)^2$...


2

A single call to partial_fit is very unlikely to get you a good fit, as it only performs one iteration of stochastic gradient descient. As stated in the docs: Internally, this method uses max_iter = 1. Therefore, it is not guaranteed that a minimum of the cost function is reached after calling it once. Matters such as objective convergence and early ...


1

I assume the "dose" $y$ is limited to $y \in [0,1]$. So in the moment you have "bunching" in your target value $y$ which you try to remove. In this case, a linear regression could lead to "overshooting" (see here for more details). So it could be beneficial to use some estimator which "restrics" $\hat{y} \in [0,1]$ as ...


0

Generally, regression models are regularized with Lasso, Ridge, or Elasticnet to improve predictive performance. Predictive performance is different than statistical significance. Predictive performance is more about the entire model than any single coefficient. Checking for the statistical significance of an individual coefficient will not impact predictive ...


0

Option include segmented regression or decision tree regression. Both of those algorithms are able learn to predict different targets values conditional on feature values.


0

The following definition makes sure that you always get symmetric functions, even when you transition to another year. Otherwise, you would get jumps between December 31 and January 1st. def gaussian(x, month, alpha): if x - month <= 365 / 2: return np.exp(- (x - month) ** 2 / (2 * alpha)) else: return np.exp(- (x - month - 365) ** ...


Top 50 recent answers are included