have a few questions about normalizing training and test data.
As far as I understand it, for certain models (like k-NN) its imperative that you normalize the training sample because similarity is based on distance and if certain explanatory variables have much higher scales than others, it will disproportionally affect the similarity measurements.
So far, while I have read some competing arguments, it seems like overall one should standardize/normalize training sets before building models.
However, I am curious if one also does this with testing data, the arguments I have read is that you can't normalize "future data" you don't know, and if you normalize the dataset before you split you are leaking "future information" into your training sample which will affect the integrity of your model.
I have tried to read what literature is out there, but have still not found an answer I fully understand, though the above make some intuitive sense.
I know it is never “that simple” but I am looking for an generalization to some degree.
What approach should I use for my supervised models?
1. Normalize then split into training and testing samples?
2. Split first then normalize each sample?
3. Split and only normalize the training sample and after test for accuracy on a non-normalized testing set?
Or will it vary substantially between models?
Thanks!
Related
I am trying to use the random forests package for classification in R.
The Variable Importance Measures listed are:
mean raw importance score of variable x for class 0
mean raw importance score of variable x for class 1
MeanDecreaseAccuracy
MeanDecreaseGini
Now I know what these "mean" as in I know their definitions. What I want to know is how to use them.
What I really want to know is what these values mean in only the context of how accurate they are, what is a good value, what is a bad value, what are the maximums and minimums, etc.
If a variable has a high MeanDecreaseAccuracy or MeanDecreaseGini does that mean it is important or unimportant? Also any information on raw scores could be useful too.
I want to know everything there is to know about these numbers that is relevant to the application of them.
An explanation that uses the words 'error', 'summation', or 'permutated' would be less helpful then a simpler explanation that didn't involve any discussion of how random forests works.
Like if I wanted someone to explain to me how to use a radio, I wouldn't expect the explanation to involve how a radio converts radio waves into sound.
An explanation that uses the words 'error', 'summation', or 'permutated'
would be less helpful then a simpler explanation that didn't involve any
discussion of how random forests works.
Like if I wanted someone to explain to me how to use a radio, I wouldn't
expect the explanation to involve how a radio converts radio waves into sound.
How would you explain what the numbers in WKRP 100.5 FM "mean" without going into the pesky technical details of wave frequencies? Frankly parameters and related performance issues with Random Forests are difficult to get your head around even if you understand some technical terms.
Here's my shot at some answers:
-mean raw importance score of variable x for class 0
-mean raw importance score of variable x for class 1
Simplifying from the Random Forest web page, raw importance score measures how much more helpful than random a particular predictor variable is in successfully classifying data.
-MeanDecreaseAccuracy
I think this is only in the R module, and I believe it measures how much inclusion of this predictor in the model reduces classification error.
-MeanDecreaseGini
Gini is defined as "inequity" when used in describing a society's distribution of income, or a measure of "node impurity" in tree-based classification. A low Gini (i.e. higher descrease in Gini) means that a particular predictor variable plays a greater role in partitioning the data into the defined classes. It's a hard one to describe without talking about the fact that data in classification trees are split at individual nodes based on values of predictors. I'm not so clear on how this translates into better performance.
For your immediate concern: higher values mean the variables are more important. This should be true for all the measures you mention.
Random forests give you pretty complex models, so it can be tricky to interpret the importance measures. If you want to easily understand what your variables are doing, don't use RFs. Use linear models or a (non-ensemble) decision tree instead.
You said:
An explanation that uses the words
'error', 'summation', or 'permutated'
would be less helpful then a simpler
explanation that didn't involve any
discussion of how random forests
works.
It's going to be awfully tough to explain much more than the above unless you dig in and learn what about random forests. I assume you're complaining about either the manual, or the section from Breiman's manual:
http://www.stat.berkeley.edu/~breiman/RandomForests/cc_home.htm#varimp
To figure out how important a variable is, they fill it with random junk ("permute" it), then see how much predictive accuracy decreases. MeanDecreaseAccuracy and MeanDecreaseGini work this way. I'm not sure what the raw importance scores are.
Interpretability is kinda tough with Random Forests. While RF is an extremely robust classifier it makes its predictions democratically. By this I mean you build hundreds or thousands of trees by taking a random subset of your variables and a random subset of your data and build a tree. Then make a prediction for all the non-selected data and save the prediction. Its robust because it deals well with the vagaries of your data set, (ie it smooths over randomly high/low values, fortuitous plots/samples, measuring the same thing 4 different ways, etc). However if you have some highly correlated variables, both may seem important as they are not both always included in each model.
One potential approach with random forests may be to help whittle down your predictors then switch to regular CART or try the PARTY package for inference based tree models. However then you must be wary about data mining issues, and making inferences about parameters.
I have a data set called Data, with 30 scaled and centered features and 1 outcome with column name OUTCOME, referred to 700k records, stored in data.table format. I computed its PCA, and observed that its first 8 components account for the 95% of the variance. I want to train a random forest in h2o, so this is what I do:
Data.pca=prcomp(Data,retx=TRUE) # compute the PCA of Data
Data.rotated=as.data.table(Data.pca$x)[,c(1:8)] # keep only first 8 components
Data.dump=cbind(Data.rotated,subset(Data,select=c(OUTCOME))) # PCA dataset plus outcomes for training
This way I have a dataset Data.dump where I have 8 features that are rotated on the PCA components, and at each record I associated its outcome.
First question: is this rational? or do I have to permute somehow the outcomes vector? or the two things are unrelated?
Then I split Data.dump in two sets, Data.train for training and Data.test for testing, all as.h2o. The I feed them to a random forest:
rf=h2o.randomForest(training_frame=Data.train,x=1:8,y=9,stopping_rounds=2,
ntrees=200,score_each_iteration=T,seed=1000000)
rf.pred=as.data.table(h2o.predict(rf,Data.test))
What happens is that rf.pred seems not so similar to the original outcomes Data.test$OUTCOME. I tried to train a neural network as well, and did not even converge, crashing R.
Second question: is it because I am carrying on some mistake from the PCA treatment? or because I badly set up the random forest? Or I am just dealing with annoying data?
I do not know where to start, as I am new to data science, but the workflow seems correct to me.
Thanks a lot in advance.
The answer to your second question (i.e. "is it the data, or did I do something wrong") is hard to know. This is why you should always try to make a baseline model first, so you have an idea of how learnable the data is.
The baseline could be h2o.glm(), and/or it could be h2o.randomForest(), but either way without the PCA step. (You didn't say if you are doing a regression or a classification, i.e. if OUTCOME is a number or a factor, but both glm and random forest will work either way.)
Going to your first question: yes, it is a reasonable thing to do, and no you don't have to (in fact, should not) involve the outcomes vector.
Another way to answer your first question is: no, it unreasonable. It may be that a random forest can see all the relations itself without needing you to use a PCA. Remember when you use a PCA to reduce the number of input dimensions you are also throwing away a bit of signal, too. You said that the 8 components only capture 95% of the variance. So you are throwing away some signal in return for having fewer inputs, which means you are optimizing for complexity at the expense of prediction quality.
By the way, concatenating the original inputs and your 8 PCA components, is another approach: you might get a better model by giving it this hint about the data. (But you might not, which is why getting some baseline models first is essential, before trying these more exotic ideas.)
Is it best to split your data into training and test sets before doing any exploratory data analysis, or do all exploration based solely on training data?
I'm working on my first full machine learning project (a recommendation system for a course capstone project) and am looking for clarification on order of operations. My rough outline is to import and clean, do exploratory analysis, train my model, and then evaluate on a test set.
I am doing exploratory data analysis now - nothing special initially, just starting with variable distributions and whatnot. But I am not sure: should I split my data into training and test sets before or after exploratory analysis?
I don't want to potentially contaminate algorithm training by inspecting the test set. However, I also don't want to miss visual trends that might reflect real signal that my poor human eye might not see after filtering, and thus potentially miss investigating an important and relevant direction while designing my algorithm.
I checked other threads, like this, but the ones I found seem to ask more about things like regularization or actual manipulation of the original data. The answers I found were mixed but prioritized splitting first. However, I don't plan to do any actual manipulation of the data before splitting it (beyond inspecting distributions and potentially doing some factor conversions).
What do you do in your own work and why?
Thanks for helping a new programmer!
To answer this question, we should remind ourselves of why, in machine learning, we split data into training, validation and testing sets (see also this question).
Training sets are used for model development. We often carefully explore this data to get ideas for feature engineering and the general structure of the machine learning model. We then train the model using the training data set.
Usually, our goal is to generate models that will perform well not only on the training data, but also on previously unseen data. Therefore, we want to avoid models that capture the peculiarities of the data we have available now rather than the general structure of the data we will see in the future ("overfitting"). To do so, we assess the quality of the models we're training by evaluating their performance on a different set of data, the validation data, and choose the model that performs best on the validation data.
Having trained our final model, we often want to have an unbiased estimate of its performance. Since we have already used the validation data in the process of model development (we chose the model that performed best on the validation data), we cannot be sure that our model will perform equally well on unseen data. So, to assess model quality, we test performance unsing a new batch of data, the testing data.
This discussion gives the answer your question: We should not use the testing (or validation) data set for exploratory data analysis. Because if we did, we would run the risk of overfitting the model to the peculiarities of the data we have, for example by engineering features that work well for the testing data. At the same time, we would lose the ability of getting an unbiased estimate of our model's performance.
I would take the problem the other way round; is it bad to use the test set ?
The objective of modeling is to end up with a model with low variance (and small bias): that's why the test set is keeping a bunch of data aside to assess how your model behaves with new data (i.e. its variance). If you use the test set during modeling you are left with nothing to do that, and you are overfitting your data.
The objective of EDA is to understand the data you're working with; the distributions of features, their relationships, their dynamics, etc ... If you leave your test set in the data, is there a risk of "overfitting" your understanding of data ? If that was the case, you would observe on say 70% of your data some properties that are not valid for the 30% remaining (test set) ... knowing that the split is random, this is impossible, or you have been extremely unlucky.
From my understanding in Machine Learning Pipeline is exploratory data analysis should be done before splitting the data into train and test.
Here are my reasons:
The data may not be cleaned in the beginning. It might have missing values, mismatch datatypes and outliers.
Need to understand every features with the target variable in the dataset. This will help to understand the importance of every features with respect to the business problem and will help to derive the additional features as well.
The data visualization will also help to get the insights information from the dataset.
Once the above operations done, then we can split the dataset into train and test. Because the features must be similar in both train and test.
I've been working with Weka for awhile now, and in my research on it, I find that a lot of code examples use test and training sets. For instance, with Discretization and Bayesian Networks,their examples are almost always shown using test and training sets. I may be missing some fundamental understanding of data processing here, but I don't understand why this seems to always be the case. I am using Discretization and Bayesian Networks in a project and for both of them, I have not used test or training sets, and do not see why I would need to either. I am performing cross validation on the BayesNet, so I am testing its accuracy. Am I misunderstanding what test and training sets are used for??? Oh and please use the simplest of terminology; I'm still not very experienced with the world of data processing.
The idea behind training and test sets is to test the generalization error. That is, if you used just one data set, you could achieve perfect accuracy by simply learning this set (this is what nearest neighbour classifiers do, IBk in Weka). In general, this is not what you want however -- the machine learning algorithm should learn the general concept behind the example data that you give it. A way of testing whether this happens is to use separate data for training and testing.
If you're using cross-validation, you're using separate training and test sets. This is simply a way of coming up with the partition of your entire data set into training and test. If you do 10 fold cross-validation for example, your entire data is partitioned into 10 sets of equal size. Nine of these are combined and used for training, the remaining one for testing. Then the process is repeated with nine different sets combined for training and so on until all the ten individual partitions will have been used for testing.
So training/test sets and cross-validation are conceptually doing the same thing, cross-validation simply takes a more rigorous approach by averaging over the entire data set.
Training data refers to the data used to "build the model".
For example, it you are using the algorithm J48 (a tree classifier) to classify instances, the training data will be used to generate the tree that will represent the "learned concept" that should be a generalization of the concept. It means that the learned rules, generated trees, the adjusted neural network, or whatever; will be able to get new (unseen) instances and classify them correctly (the "learned concept" does not depends on the training data).
The test sets are a percentage of the data that will be used to test whether the model has learned the concept properly (it is independent of the training data).
In WEKA you can run an execution splitting your data set into trainig data (to build the tree in the case of J48) and test data (to test the model in order to determine that the concept has been learned). For example, you can use 60% of the data for training and 40% for testing (determine how much data is needed for training and testing is one of the key problems of data mining).
But I would recommend you to have a quick look to cross-validation, that is a robust testing method that is implemented in WEKA. It has been explained quite well here:
https://stackoverflow.com/a/10539247/1565171
If you have more questions just leave a comment.
I am trying to use the random forests package for classification in R.
The Variable Importance Measures listed are:
mean raw importance score of variable x for class 0
mean raw importance score of variable x for class 1
MeanDecreaseAccuracy
MeanDecreaseGini
Now I know what these "mean" as in I know their definitions. What I want to know is how to use them.
What I really want to know is what these values mean in only the context of how accurate they are, what is a good value, what is a bad value, what are the maximums and minimums, etc.
If a variable has a high MeanDecreaseAccuracy or MeanDecreaseGini does that mean it is important or unimportant? Also any information on raw scores could be useful too.
I want to know everything there is to know about these numbers that is relevant to the application of them.
An explanation that uses the words 'error', 'summation', or 'permutated' would be less helpful then a simpler explanation that didn't involve any discussion of how random forests works.
Like if I wanted someone to explain to me how to use a radio, I wouldn't expect the explanation to involve how a radio converts radio waves into sound.
An explanation that uses the words 'error', 'summation', or 'permutated'
would be less helpful then a simpler explanation that didn't involve any
discussion of how random forests works.
Like if I wanted someone to explain to me how to use a radio, I wouldn't
expect the explanation to involve how a radio converts radio waves into sound.
How would you explain what the numbers in WKRP 100.5 FM "mean" without going into the pesky technical details of wave frequencies? Frankly parameters and related performance issues with Random Forests are difficult to get your head around even if you understand some technical terms.
Here's my shot at some answers:
-mean raw importance score of variable x for class 0
-mean raw importance score of variable x for class 1
Simplifying from the Random Forest web page, raw importance score measures how much more helpful than random a particular predictor variable is in successfully classifying data.
-MeanDecreaseAccuracy
I think this is only in the R module, and I believe it measures how much inclusion of this predictor in the model reduces classification error.
-MeanDecreaseGini
Gini is defined as "inequity" when used in describing a society's distribution of income, or a measure of "node impurity" in tree-based classification. A low Gini (i.e. higher descrease in Gini) means that a particular predictor variable plays a greater role in partitioning the data into the defined classes. It's a hard one to describe without talking about the fact that data in classification trees are split at individual nodes based on values of predictors. I'm not so clear on how this translates into better performance.
For your immediate concern: higher values mean the variables are more important. This should be true for all the measures you mention.
Random forests give you pretty complex models, so it can be tricky to interpret the importance measures. If you want to easily understand what your variables are doing, don't use RFs. Use linear models or a (non-ensemble) decision tree instead.
You said:
An explanation that uses the words
'error', 'summation', or 'permutated'
would be less helpful then a simpler
explanation that didn't involve any
discussion of how random forests
works.
It's going to be awfully tough to explain much more than the above unless you dig in and learn what about random forests. I assume you're complaining about either the manual, or the section from Breiman's manual:
http://www.stat.berkeley.edu/~breiman/RandomForests/cc_home.htm#varimp
To figure out how important a variable is, they fill it with random junk ("permute" it), then see how much predictive accuracy decreases. MeanDecreaseAccuracy and MeanDecreaseGini work this way. I'm not sure what the raw importance scores are.
Interpretability is kinda tough with Random Forests. While RF is an extremely robust classifier it makes its predictions democratically. By this I mean you build hundreds or thousands of trees by taking a random subset of your variables and a random subset of your data and build a tree. Then make a prediction for all the non-selected data and save the prediction. Its robust because it deals well with the vagaries of your data set, (ie it smooths over randomly high/low values, fortuitous plots/samples, measuring the same thing 4 different ways, etc). However if you have some highly correlated variables, both may seem important as they are not both always included in each model.
One potential approach with random forests may be to help whittle down your predictors then switch to regular CART or try the PARTY package for inference based tree models. However then you must be wary about data mining issues, and making inferences about parameters.