Pushing the docs to dev/ for branch: main, commit 74e23aab06d2456401470c020034142e766b10b4

Author: sklearn-ci

dev/.buildinfo

@@ -1,4 +1,4 @@
 # Sphinx build info version 1
 # This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done.
-config: 14fe89ee948c902f97edbd045f4b1a63
+config: a28be30aafd91c9eaba1819985c6c865
 tags: 645f666f9bcd5a90fca523b33c5a78b7

dev/_downloads/07fcc19ba03226cd3d83d4e40ec44385/auto_examples_python.zip

@@ -76,7 +76,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "With the heteroscedastic Normal distributed target, we observe that the\nvariance of the noise is increasing when the value of the feature `x` is\nincreasing.\n\nWith the asymmetric Pareto distributed target, we observe that the positive\nresiduals are bounded.\n\nThese types of noisy targets make the estimation via\n:class:`~sklearn.linear_model.LinearRegression` less efficient, i.e. we need\nmore data to get stable results and, in addition, large outliers can have a\nhuge impact on the fitted coefficients. (Stated otherwise: in a setting with\nconstant variance, ordinary least squares estimators converge much faster to\nthe *true* coefficients with increasing sample size.)\n\nIn this asymmetric setting, the median or different quantiles give additional\ninsights. On top of that, median estimation is much more robust to outliers\nand heavy tailed distributions. But note that extreme quantiles are estimated\nby very view data points. 95% quantile are more or less estimated by the 5%\nlargest values and thus also a bit sensitive outliers.\n\nIn the remainder of this tutorial, we will show how\n:class:`~sklearn.linear_model.QuantileRegressor` can be used in practice and\ngive the intuition into the properties of the fitted models. Finally,\nwe will compare the both :class:`~sklearn.linear_model.QuantileRegressor`\nand :class:`~sklearn.linear_model.LinearRegression`.\n\n## Fitting a `QuantileRegressor`\n\nIn this section, we want to estimate the conditional median as well as\na low and high quantile fixed at 5% and 95%, respectively. Thus, we will get\nthree linear models, one for each quantile.\n\nWe will use the quantiles at 5% and 95% to find the outliers in the training\nsample beyond the central 90% interval.\n\n"
+        "With the heteroscedastic Normal distributed target, we observe that the\nvariance of the noise is increasing when the value of the feature `x` is\nincreasing.\n\nWith the asymmetric Pareto distributed target, we observe that the positive\nresiduals are bounded.\n\nThese types of noisy targets make the estimation via\n:class:`~sklearn.linear_model.LinearRegression` less efficient, i.e. we need\nmore data to get stable results and, in addition, large outliers can have a\nhuge impact on the fitted coefficients. (Stated otherwise: in a setting with\nconstant variance, ordinary least squares estimators converge much faster to\nthe *true* coefficients with increasing sample size.)\n\nIn this asymmetric setting, the median or different quantiles give additional\ninsights. On top of that, median estimation is much more robust to outliers\nand heavy tailed distributions. But note that extreme quantiles are estimated\nby very few data points. 95% quantile are more or less estimated by the 5%\nlargest values and thus also a bit sensitive outliers.\n\nIn the remainder of this tutorial, we will show how\n:class:`~sklearn.linear_model.QuantileRegressor` can be used in practice and\ngive the intuition into the properties of the fitted models. Finally,\nwe will compare the both :class:`~sklearn.linear_model.QuantileRegressor`\nand :class:`~sklearn.linear_model.LinearRegression`.\n\n## Fitting a `QuantileRegressor`\n\nIn this section, we want to estimate the conditional median as well as\na low and high quantile fixed at 5% and 95%, respectively. Thus, we will get\nthree linear models, one for each quantile.\n\nWe will use the quantiles at 5% and 95% to find the outliers in the training\nsample beyond the central 90% interval.\n\n"
       ]
     },
     {
@@ -170,7 +170,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "On the training set, we see that MAE is lower for\n:class:`~sklearn.linear_model.QuantileRegressor` than\n:class:`~sklearn.linear_model.LinearRegression`. In contrast to that, MSE is\nlower for :class:`~sklearn.linear_model.LinearRegression` than\n:class:`~sklearn.linear_model.QuantileRegressor`. These results confirms that\nMAE is the loss minimized by :class:`~sklearn.linear_model.QuantileRegressor`\nwhile MSE is the loss minimized\n:class:`~sklearn.linear_model.LinearRegression`.\n\nWe can make a similar evaluation but looking at the test error obtained by\ncross-validation.\n\n"
+        "On the training set, we see that MAE is lower for\n:class:`~sklearn.linear_model.QuantileRegressor` than\n:class:`~sklearn.linear_model.LinearRegression`. In contrast to that, MSE is\nlower for :class:`~sklearn.linear_model.LinearRegression` than\n:class:`~sklearn.linear_model.QuantileRegressor`. These results confirms that\nMAE is the loss minimized by :class:`~sklearn.linear_model.QuantileRegressor`\nwhile MSE is the loss minimized\n:class:`~sklearn.linear_model.LinearRegression`.\n\nWe can make a similar evaluation by looking at the test error obtained by\ncross-validation.\n\n"
       ]
     },
     {

dev/_downloads/10754505339af88c10b8a48127535a4a/plot_quantile_regression.ipynb

@@ -93,7 +93,7 @@
 # In this asymmetric setting, the median or different quantiles give additional
 # insights. On top of that, median estimation is much more robust to outliers
 # and heavy tailed distributions. But note that extreme quantiles are estimated
-# by very view data points. 95% quantile are more or less estimated by the 5%
+# by very few data points. 95% quantile are more or less estimated by the 5%
 # largest values and thus also a bit sensitive outliers.
 #
 # In the remainder of this tutorial, we will show how
@@ -281,7 +281,7 @@
 # while MSE is the loss minimized
 # :class:`~sklearn.linear_model.LinearRegression`.
 #
-# We can make a similar evaluation but looking at the test error obtained by
+# We can make a similar evaluation by looking at the test error obtained by
 # cross-validation.
 from sklearn.model_selection import cross_validate