Merge pull request #99916 from vmagelo/automl-sdk-fixes

Changes to articles prompted by SDK work.

Author: PRMerger20

articles/machine-learning/concept-automated-ml.md

@@ -24,7 +24,7 @@ With automated machine learning, you'll accelerate the time it takes to get prod
 
 Automated ML democratizes the machine learning model development process, and empowers its users, no matter their data science expertise, to identify an end-to-end machine learning pipeline for any problem.
 
-Data scientists, analysts and developers across industries can use automated ML to:
+Data scientists, analysts, and developers across industries can use automated ML to:
 
 + Implement machine learning solutions without extensive programming knowledge
 + Save time and resources
@@ -88,14 +88,14 @@ In every automated machine learning experiment, your data is automatically scale
 | [MaxAbsScaler](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MaxAbsScaler.html#sklearn.preprocessing.MaxAbsScaler) |Scale each feature by its maximum absolute value |
 | [RobustScalar](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.RobustScaler.html) |This Scaler features by their quantile range |
 | [PCA](https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.PCA.html) |Linear dimensionality reduction using Singular Value Decomposition of the data to project it to a lower dimensional space |
-| [TruncatedSVDWrapper](https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.TruncatedSVD.html) |This transformer performs linear dimensionality reduction by means of truncated singular value decomposition (SVD). Contrary to PCA, this estimator does not center the data before computing the singular value decomposition. This means it can work with scipy.sparse matrices efficiently |
-| [SparseNormalizer](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.Normalizer.html) | Each sample (that is, each row of the data matrix) with at least one non-zero component is re-scaled independently of other samples so that its norm (l1 or l2) equals one |
+| [TruncatedSVDWrapper](https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.TruncatedSVD.html) |This transformer performs linear dimensionality reduction by means of truncated singular value decomposition (SVD). Contrary to PCA, this estimator does not center the data before computing the singular value decomposition, which means it can work with scipy.sparse matrices efficiently |
+| [SparseNormalizer](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.Normalizer.html) | Each sample (that is, each row of the data matrix) with at least one non-zero component is rescaled independently of other samples so that its norm (l1 or l2) equals one |
 
 ### Advanced preprocessing: optional featurization
 
 Additional advanced preprocessing and featurization are also available, such as missing values imputation, encoding, and transforms. [Learn more about what featurization is included](how-to-create-portal-experiments.md#preprocess). Enable this setting with:
 
-+ Azure Machine Learning studio : Selecting the **View featurization settings** in the **Configuration Run** section [with these steps](how-to-create-portal-experiments.md).
++ Azure Machine Learning studio: Selecting the **View featurization settings** in the **Configuration Run** section [with these steps](how-to-create-portal-experiments.md).
 
 + Python SDK: Specifying `"feauturization": auto' / 'off' / FeaturizationConfig` for the [`AutoMLConfig` class](/python/api/azureml-train-automl-client/azureml.train.automl.automlconfig.automlconfig).
 
@@ -107,7 +107,7 @@ The best way to prevent over-fitting is to follow ML best-practices including:
 
 * Using more training data, and eliminating statistical bias
 * Preventing target leakage
-* Using less features
+* Using fewer features
 * **Regularization and hyperparameter optimization**
 * **Model complexity limitations**
 * **Cross-validation**
@@ -118,15 +118,15 @@ In the context of automated ML, the first three items above are **best-practices
 
 Using **more data** is the simplest and best possible way to prevent over-fitting, and as an added bonus typically increases accuracy. When you use more data, it becomes harder for the model to memorize exact patterns, and it is forced to reach solutions that are more flexible to accommodate more conditions. It's also important to recognize **statistical bias**, to ensure your training data doesn't include isolated patterns that won't exist in live-prediction data. This scenario can be difficult to solve, because there may not be over-fitting between your train and test sets, but there may be over-fitting present when compared to live test data.
 
-Target leakage is a similar issue, where you may not see over-fitting between train/test sets, but rather it appears at prediction-time. Target leakage occurs when your model "cheats" during training by having access to data that it shouldn't normally have at prediction-time. For example, if your problem is to predict on Monday what a commodity price will be on Friday, but one of your features accidentally included data from Thursdays, that would be data the model won't have at prediction-time since it cannot see into the future. Target leakage is an easy mistake to miss, but is often characterized by abnormally-high accuracy for your problem. If you are attempting to predict stock price and trained a model at 95% accuracy, there is very likely target leakage somewhere in your features.
+Target leakage is a similar issue, where you may not see over-fitting between train/test sets, but rather it appears at prediction-time. Target leakage occurs when your model "cheats" during training by having access to data that it shouldn't normally have at prediction-time. For example, if your problem is to predict on Monday what a commodity price will be on Friday, but one of your features accidentally included data from Thursdays, that would be data the model won't have at prediction-time since it cannot see into the future. Target leakage is an easy mistake to miss, but is often characterized by abnormally high accuracy for your problem. If you are attempting to predict stock price and trained a model at 95% accuracy, there is likely target leakage somewhere in your features.
 
 Removing features can also help with over-fitting by preventing the model from having too many fields to use to memorize specific patterns, thus causing it to be more flexible. It can be difficult to measure quantitatively, but if you can remove features and retain the same accuracy, you have likely made the model more flexible and have reduced the risk of over-fitting.
 
 ### Best-practices automated ML implements
 
 Regularization is the process of minimizing a cost function to penalize complex and over-fitted models. There are different types of regularization functions, but in general they all penalize model coefficient size, variance, and complexity. Automated ML uses L1 (Lasso), L2 (Ridge), and ElasticNet (L1 and L2 simultaneously) in different combinations with different model hyperparameter settings that control over-fitting. In simple terms, automated ML will vary how much a model is regulated and choose the best result.
 
-Automated ML also implements explicit model complexity limitations to prevent over-fitting. In most cases this is specifically for decision tree or forest algorithms, where individual tree max-depth is limited, and the total number of trees used in forest or ensemble techniques are limited.
+Automated ML also implements explicit model complexity limitations to prevent over-fitting. In most cases this implementation is specifically for decision tree or forest algorithms, where individual tree max-depth is limited, and the total number of trees used in forest or ensemble techniques are limited.
 
 Cross-validation (CV) is the process of taking many subsets of your full training data and training a model on each subset. The idea is that a model could get "lucky" and have great accuracy with one subset, but by using many subsets the model won't achieve this high accuracy every time. When doing CV, you provide a validation holdout dataset, specify your CV folds (number of subsets) and automated ML will train your model and tune hyperparameters to minimize error on your validation set. One CV fold could be over-fit, but by using many of them it reduces the probability that your final model is over-fit. The tradeoff is that CV does result in longer training times and thus greater cost, because instead of training a model once, you train it once for each *n* CV subsets.
 
@@ -148,7 +148,7 @@ Considering model **A**, there is a common misconception that if test accuracy o
 
 When comparing models **A** and **B**, model **A** is a better model because it has higher test accuracy, and although the test accuracy is slightly lower at 95%, it is not a significant difference that suggests over-fitting is present. You wouldn't choose model **B** simply because the train and test accuracies are closer together.
 
-Model **C** represents a clear case of over-fitting; the training accuracy is very high but the test accuracy isn't anywhere near as high. This distinction is somewhat subjective, but comes from knowledge of your problem and data, and what magnitudes of error are acceptable. 
+Model **C** represents a clear case of over-fitting; the training accuracy is very high but the test accuracy isn't anywhere near as high. This distinction is subjective, but comes from knowledge of your problem and data, and what magnitudes of error are acceptable. 
 
 ## Time-series forecasting
 
@@ -172,7 +172,7 @@ Automated machine learning supports ensemble models, which are enabled by defaul
 * **Voting**: predicts based on the weighted average of predicted class probabilities (for classification tasks) or predicted regression targets (for regression tasks).
 * **Stacking**: stacking combines heterogenous models and trains a meta-model based on the output from the individual models. The current default meta-models are LogisticRegression for classification tasks and ElasticNet for regression/forecasting tasks.
 
-The [Caruana ensemble selection algorithm](http://www.niculescu-mizil.org/papers/shotgun.icml04.revised.rev2.pdf) with sorted ensemble initialization is used to decide which models to use within the ensemble. At a high level, this algorithm initializes the ensemble with up to 5 models with the best individual scores, and verifies that these models are within 5% threshold of the best score to avoid a poor initial ensemble. Then for each ensemble iteration, a new model is added to the existing ensemble and the resulting score is calculated. If a new model improved the existing ensemble score, the ensemble is updated to include the new model.
+The [Caruana ensemble selection algorithm](http://www.niculescu-mizil.org/papers/shotgun.icml04.revised.rev2.pdf) with sorted ensemble initialization is used to decide which models to use within the ensemble. At a high level, this algorithm initializes the ensemble with up to five models with the best individual scores, and verifies that these models are within 5% threshold of the best score to avoid a poor initial ensemble. Then for each ensemble iteration, a new model is added to the existing ensemble and the resulting score is calculated. If a new model improved the existing ensemble score, the ensemble is updated to include the new model.
 
 See the [how-to](how-to-configure-auto-train.md#ensemble) for changing default ensemble settings in automated machine learning.
 
@@ -182,7 +182,7 @@ Imbalanced data is commonly found in data for machine learning classification sc
 
 As part of its goal of simplifying the machine learning workflow, automated ML has built in capabilities to help deal with imbalanced data such as, 
 
-- A **weight column**: automated ML supports a weighted column as input, causing rows in the data to be weighted up or down, which can make a class more or less “important”. See this [notebook example](https://github.com/Azure/MachineLearningNotebooks/blob/master/how-to-use-azureml/automated-machine-learning/sample-weight/auto-ml-sample-weight.ipynb) 
+- A **weight column**: automated ML supports a weighted column as input, causing rows in the data to be weighted up or down, which can make a class more or less “important”.
 
 - The algorithms used by automated ML can properly handle imbalance of up to 20:1, meaning the most common class can have 20 times more rows in the data than the least common class.
 
@@ -208,7 +208,7 @@ The following techniques are additional options to handle imbalanced data outsid
 
 ## Use with ONNX in C# apps
 
-With Azure Machine Learning, you can use automated ML to build a Python model and have it converted to the ONNX format. The ONNX runtime supports  C#, so you can use the model built automatically in your C# apps without any need for recoding or any of the network latencies that REST endpoints introduce. Try an example of this flow [in this Jupyter notebook](https://github.com/Azure/MachineLearningNotebooks/blob/master/how-to-use-azureml/automated-machine-learning/classification-with-onnx/auto-ml-classification-with-onnx.ipynb).
+With Azure Machine Learning, you can use automated ML to build a Python model and have it converted to the ONNX format. The ONNX runtime supports  C#, so you can use the model built automatically in your C# apps without any need for recoding or any of the network latencies that REST endpoints introduce. Try an example of this flow [in this Jupyter notebook](https://github.com/Azure/MachineLearningNotebooks/blob/master/how-to-use-azureml/automated-machine-learning/classification-bank-marketing-all-features/auto-ml-classification-bank-marketing-all-features.ipynb).
 
 ## Automated ML across Microsoft