WEKA cross validation discretization - weka

I'm trying to improve the accuracy of my WEKA model by applying an unsupervised discretize filter. I need to decided on the number of bins and whether equal frequency binning should be used. Normally, I would optimize this using a training set.
However, how do I determine the bin size and whether equal frequency binning should be used when using cross-validation? My initial idea was to use the accuracy result of the classifier in multiple cross-validation tests to find the optimal bin size. However, isn't it wrong, despite using cross-validation, to use this same set to also test the accuracy of the model, because I then have an overfitted model? What would then be a correct way of determining the bin sizes?
I also tried the supervized discretize filter to determine the bin sizes, however this results in only in single bins. Does this mean that my data is too random and therefore cannot be clustered into multiple bins?

Yes, you are correct in both your idea and your concerns for the first issue.
What you are trying to do is Parameter Optimization. This term is usually used when you try to optimize the parameters of your classifier, e.g., the number of trees for the Random Forest or the C parameter for SVMs. But you can apply it as well to pre-processing steps and filters.
What you have to do in this case is a nested cross-validation. (You should check https://stats.stackexchange.com/ for more information, for example here or here). It is important that the final classifier, including all pre-processing steps like binning and such, has never seen the test set, only the training set. This is the outer cross-validation.
For each fold of the outer cross-validation, you need to do an inner cross-validation on the training set to determine the optimal parameters for your model.
I'll try to "visualize" it on a simple 2-fold cross-validation
Data set
########################################
Split for outer cross-validation (2-fold)
#################### ####################
training set test set
Split for inner cross-validation
########## ##########
training test
Evaluate parameters
########## ##########
build with evaluated
bin size 5 acc 70%
bin size 10 acc 80%
bin size 20 acc 75%
...
=> optimal bin size: 10
Outer cross-validation (2-fold)
#################### ####################
training set test set
apply bin size 10
train model evaluate model
Parameter optimization can be very exhausting. If you have 3 parameters with 10 possible parameter values each, that makes 10x10x10=1000 parameter combinations you need to evaluate for each outer fold.
This is a topic of machine learning by itself, because you can do everything from the naive grid search to evolutionary search here. Sometimes you can use heuristics. But you need to do some kind of parameter optimization every time.
As for your second question: This is really hard to tell without seeing your data. But you should post that as a separate question anyway.

Related

Clarification re Principle Component Analysis

I do understand the principle component analysis. I know how to do it and what it actually does. I have applied PCA and my best result has shown to be two components. I do understand that each of my inputs are now contributing partially in each component. What I do not understand is how to feed the result of PCA (in my case 2 components ) to a machine learning model?
How do we input them?
For example when I want to run a NN on my features, I just can navigate to where they are stored and import them, but my PCA analysis has been run in SPSS and all it shows me is the contribution of my features on each component.
What should I import to my NN model?
PCA is a method of feature extraction, which is used to avoid the problem of co-linearity. For example, if several variables are highly correlated because "they measure the same thing", then PCA can extract a measure of "that thing" (technically: a component), which is called a score. Your data set of, say, 100 measured variables may reduce to, say, 10 significant components. Then you can use the scores your test persons have achieved in those 10 components to do for example a multi-dimensional regression, a cluster analysis or a discriminance analysis. This will result in more valid results than performing the analysis directly on the 100 variables.
So the procedure is to sort the eigenvalues (and -vectors) by size, identify the number of significant components p (e.g., by scree-plot), set up the projection matrix F (eigenvectors corresponding to the largest q eigenvalues in columns) and multiply it with the data matrix D. This will give you the score matrix C (dimension n times q, with n the number of test persons), which you can use as input for whatever method you want to use next.

When training a single batch, is iteration of examples necessary (optimal) in python code?

Say I have one batch that I want to train my model on. Do I simply run tf.Session()'s sess.run(batch) once, or do I have to iterate through all of the batch's examples with a loop in the session? I'm looking for the optimal way to iterate/update the training ops, such as loss. I thought tensorflow would handle it itself, especially in the cases where tf.nn.dynamic_rnn() takes in a batch dimension for listing the examples. I thought, perhaps naively, that a for loop in the python code would be the inefficient method of updating the loss. I am using tf.losses.mean_squared_error(batch) for a regression problem.
My regression problem is given two lists of word vectors (300d each), and determines the similarity between the two lists on a continuous scale from [0, 5]. My supervised model is Deepmind's Differential Neural Computer (DNC). The problem is I do not believe it is learning anything. this is due to the fact that the all of the output from the model is centered around 0 and even negative. I do not know how it could possibly be negative given no negative labels provided. I only call sess.run(loss) for the single batch, I do not create a python loop to iterate through it.
So, what is the most efficient way to iterate the training of a model and how do people go about it? Do they really use python loops to do multiple calls to sess.run(loss) (this was done in the training file example for DNC, and I have seen it in other examples as well). I am certain I get the final loss from the below process, but I am uncertain if the model has actually been trained entirely just because the loss was processed in one go. I also do not understand the point of update_ops returned by some functions, and am uncertain if they are necessary to ensure the model has been trained.
Example of what I mean by processing a batch's loss once:
# assume the model has been defined prior through batch_output_logits
train_loss = tf.losses.mean_squared_error(labels=target,
predictions=batch_output_logits)
with tf.Session() as sess:
sess.run(init_op) # pseudo code, unnecessary for question
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
# is this the entire batch's loss && model has been trained for that batch?
loss_np = sess.run(train_step, train_loss)
coord.request_stop()
coord.join(threads)
Any input on why I am receiving negative values when the labels are in the range [0, 5] is welcomed as well(general abstract answers for this are fine, because its not the main focus). I am thinking of attempting to create a piece-wise function, if possible, for my loss, so that for any values out of bounds face a rapidly growing exponential loss function. Uncertain how to implement, or if it would even work.
Code is currently private. Once allowed, I will make the repo public.
To run DNC model, go to the project/ directory and run python -m src.main. If there are errors you encounter feel free to let me know.
This model depends upon Tensorflow r1.2, most recent Sonnet, and NLTK's punkt for Tokenizing sentences in sts_handler.py and tests/*.
In a regression model, the network calculates the model output based on the randomly initialized values for your model parameters. That's why you're seeing negative values here; you haven't trained your model enough for it to learn that your values are only between 0 and 5.
Unless I'm missing something, you are only calculating the loss, but you aren't actually training the model. You should probably be calling sess.run(optimizer) on an optimizer, not on your loss function.
You probably need to train your model for multiple epochs (training your model for one epoch = training your model once on the entire dataset).
Batches are used because it is more computationally efficient to train your model on a batch than it is to train it on a single example. However, your data seems to be small enough that you won't have that problem. As such, I would recommend reducing your batch size to as low as possible. As a general rule, you get better training from a smaller batch size, at the cost of added computation.
If you post all of your code, I can take a look.

Imbalance between errors in data summary and tree visualization in Weka

I tried to run a simple classification on the iris.arff dataset in Weka, using the J48 algorithm. I used cross-validation with 10 folds and - if I'm not wrong - all the default settings for J48.
The result is a 96% accuracy with 6 incorrectly classified instances.
Here's my question: according to this the second number in the tree visualization is the number of the wrongly classified instances in each leaf, but then why their sum isn't 6 but 3?
EDIT: running the algorithm with different test options I obtain different results in terms of accuracy (and therefore number of errors), but when I visualize the tree I get always the same tree with the same 3 errors. I still can't explain why.
The second number in the tree visualization is not the number of the wrongly classified instances in each leaf - it's the total weight of those wrongly classified instances.
Did you, by any chance, weigh some of those instances with 0.5 instead of 1?
Another option is that you are actually executing two different models. One where you use the full training set to build the classifier (classifier.buildClassifier(instances)) and another one where you run Cross-validation (eval.crossValidateModel(...)) with 10 train/test folds. The first model will produce the visualised tree with less errors (larger trainingset) while the second model from CV produces the output statistics with more errors. This would explain why you get different stats when changing the test set but still the same tree that is built on the full set.
For the record: if you train (and visualise) the tree with the full dataset, you will appear to have less errors, but your model will actually be overfitted and the obtained performance measures will probably not be realistic. As such, your results from CV are much more useful and you should visualise the tree from that model.

Recursive Feature Elimination CV in Sklearn changes when I remove features

I'm using the RFECV module in sklearn to find the optimal number of features to yield the highest Cross validation on 2 folds. I am using a ridge regressor as my estimator.
rfecv = RFECV(estimator=ridge,step=1, cv=KFold(n_splits=2))
rfecv.fit(df, y)
I have 5 features in my dataset that I have standardized using the standardscaler.
I'll run the RFECV on my data, and it'll say that 2 features is optimal. But when I remove one of the features with the lowest regression coefficient and rerun the RFECV, it now says that 3 features is optimal.
When I progress through all features one at a time (as the recursive should do) I find that 3 is in fact the optimal.
I've tested this with other datasets, and have found that the optimal number of features changes as I remove features one at a time and rerun RFECV.
I might be missing something, but isn't that what RFECV is supposed to solve?
Any additional insights on RFECV is appreciated.
This makes sense actually. RFECV is recommending a certain number of features based on the available data. When you remove the feature you change the scoring range.
from the docs:
# Determine the number of subsets of features by fitting across
# the train folds and choosing the "features_to_select" parameter
# that gives the least averaged error across all folds.
...
n_features_to_select = max(
n_features - (np.argmax(scores) * step),
n_features_to_select)
n_features_to_select is used to determine how many features should be used in RFE for any particular iteration (within/under-the-hood of RFECV).
rfe = RFE(estimator=self.estimator,
n_features_to_select=n_features_to_select,
step=self.step, verbose=self.verbose)
And so this is directly connected to the number of features you include in your initial rfecv.fit() step.
Also, removing the feature with the lowest regression coefficient is not the best way to trim features. The coefficient is a reflection of its impact on the dependent variable not necessarily the model's accuracy.

Regression Tree Forest in Weka

I'm using Weka and would like to perform regression with random forests. Specifically, I have a dataset:
Feature1,Feature2,...,FeatureN,Class
1.0,X,...,1.4,Good
1.2,Y,...,1.5,Good
1.2,F,...,1.6,Bad
1.1,R,...,1.5,Great
0.9,J,...,1.1,Horrible
0.5,K,...,1.5,Terrific
.
.
.
Rather than learning to predict the most likely class, I want to learn the probability distribution over the classes for a given feature vector. My intuition is that using just the RandomForest model in Weka would not be appropriate, since it would be attempting to minimize its absolute error (maximum likelihood) rather than its squared error (conditional probability distribution). Is that intuition right? Is there a better model to be using if I want to perform regression rather than classification?
Edit: I'm actually thinking now that in fact it may not be a problem. Presumably, classifiers are learning the conditional probability P(Class | Feature1,...,FeatureN) and the resulting classification is just finding the c in Class that maximizes that probability distribution. Therefore, a RandomForest classifier should be able to give me the conditional probability distribution. I just had to think about it some more. If that's wrong, please correct me.
If you want to predict the probabilities for each class explicitly, you need different input data. That is, you would need to replace the value to predict. Instead of one data set with the class label, you would need n data sets (for n different labels) with aggregated data for each unique feature vector. Your data would look something like
Feature1,...,Good
1.0,...,0.5
0.3,...,1.0
and
Feature1,...,Bad
1.0,...,0.8
0.3,...,0.1
and so on. You would need to learn one model for each class and run them separately on any data to be classified. That is, for each label you learn a model to predict a number that is the probability of being in that class, given a feature vector.
If you don't need the probabilities to be predicted explicitly, have a look at the Bayesian classifiers in Weka, which make use of probabilities in the models that they learn.