We Keep Coding

multiclass CV in XGboost (python) - some classes not in train/validation sub-groups

I am working with an XGboost model in python, with a large dataset comprised of embeddings (x) and corresponding labels (y), I have about 30000 samples.
The data is very imbalanced, with 8 different classes of labels.
I am attempting to perform hyperparameter tuning (using RandomizedSearchCV).
For some of the CV folds I get an error:
ValueError: Invalid classes inferred from unique values of `y`. Expected: [0 1 2 3 4 5 6 7], got [0 1 2 3 5 6 7 8].
Due to the different splitting each time (using stratified split), some splits do not have all the labels in both groups.
I searched the web a lot and couldn't find anything in this exact context, even though I imagine this should be a major issue for many imbalanced multiclass classifications.
My code:
y = y.values.astype(int)
le = LabelEncoder()
y = le.fit_transform(y)
xgb_base = XGBClassifier(objective='multi:softprob', learning_rate=LR)
cv = StratifiedGroupKFold(n_splits=NUM_CV)
# Create the random search Random Forest
xgb_random = RandomizedSearchCV(estimator=xgb_base, param_distributions=xgb_grid,
n_iter=NUM_ITER, cv=cv, verbose=2,
random_state=1)
# Fit the random search model
xgb_random.fit(X, y, groups=groups)
# Get the optimal parameters
xgb_random.best_params_
print(xgb_random.best_params_)

This is not a bug or error. Use StratifiedCV and see if that helps.
Why this is happening:
Let us suppose you have 3 classes and 5 samples as [0,1,0,1,2]. So if even if you split it into 2 groups i.e k=2, either train or test won't have the class == 2. This is happening with your case.
If you have K > minimum number of samples per class, you'll definitely have this problem. If not, then StratifiedKFold can help. It'll split the data in a manner so that each split has almost the same distribution of classes.
On a broader note, if you can, then drop the classes not required or try merging two or more classes if you can.
Check this link to see the difference between different KFold types

Building ML classifier with imbalanced data

I have a dataset with 1400 obs and 19 columns. The Target variable has values 1 (value that I am most interested in) and 0. The distribution of classes shows imbalance (70:30).
Using the code below I am getting weird values (all 1s). I am not figuring out if this is due to a problem of overfitting/imbalance data or to feature selection (I used Pearson correlation since all values are numeric/boolean).
I am thinking that the steps followed are wrong.
import numpy as np
import math
import sklearn.metrics as metrics
from sklearn.metrics import f1_score
y = df['Label']
X = df.drop('Label',axis=1)
def create_cv(X,y):
if type(X)!=np.ndarray:
X=X.values
y=y.values
test_size=1/5
proportion_of_true=y[y==1].shape[0]/y.shape[0]
num_test_samples=math.ceil(y.shape[0]*test_size)
num_test_true_labels=math.floor(num_test_samples*proportion_of_true)
num_test_false_labels=math.floor(num_test_samples-num_test_true_labels)
y_test=np.concatenate([y[y==0][:num_test_false_labels],y[y==1][:num_test_true_labels]])
y_train=np.concatenate([y[y==0][num_test_false_labels:],y[y==1][num_test_true_labels:]])
X_test=np.concatenate([X[y==0][:num_test_false_labels] ,X[y==1][:num_test_true_labels]],axis=0)
X_train=np.concatenate([X[y==0][num_test_false_labels:],X[y==1][num_test_true_labels:]],axis=0)
return X_train,X_test,y_train,y_test
X_train,X_test,y_train,y_test=create_cv(X,y)
X_train,X_crossv,y_train,y_crossv=create_cv(X_train,y_train)
tree = DecisionTreeClassifier(max_depth = 5)
tree.fit(X_train, y_train)
y_predict_test = tree.predict(X_test)
print(classification_report(y_test, y_predict_test))
f1_score(y_test, y_predict_test)
Output:
precision recall f1-score support
0 1.00 1.00 1.00 24
1 1.00 1.00 1.00 70
accuracy 1.00 94
macro avg 1.00 1.00 1.00 94
weighted avg 1.00 1.00 1.00 94
Has anyone experienced similar issues in building a classifier when data has imbalance, using CV and/or under sampling? Happy to share the whole dataset, in case you might want to replicate the output.
What I would like to ask you for some clear answer to follow that can show me the steps and what I am doing wrong.
I know that, to reduce overfitting and work with balance data, there are some methods such as random sampling (over/under), SMOTE, CV. My idea is
Split the data on train/test taking into account imbalance
Perform CV on trains set
Apply undersampling only on a test fold
After the model has been chosen with the help of CV, undersample the train set and train the classifier
Estimate the performance on the untouched test set
(f1-score)
as also outlined in this question: CV and under sampling on a test fold .
I think the steps above should make sense, but happy to receive any feedback that you might have on this.

When you have imbalanced data you have to perform stratification. The usual way is to oversample the class that has less values.
Another option is to train your algorithm with less data. If you have a good dataset that should not be a problem. In this case you grab first the samples from the less represented class use the size of the set to compute how many samples to get from the other class:
This code may help you split your dataset that way:
def split_dataset(dataset: pd.DataFrame, train_share=0.8):
"""Splits the dataset into training and test sets"""
all_idx = range(len(dataset))
train_count = int(len(all_idx) * train_share)
train_idx = random.sample(all_idx, train_count)
test_idx = list(set(all_idx).difference(set(train_idx)))
train = dataset.iloc[train_idx]
test = dataset.iloc[test_idx]
return train, test
def split_dataset_stratified(dataset, target_attr, positive_class, train_share=0.8):
"""Splits the dataset as in `split_dataset` but with stratification"""
data_pos = dataset[dataset[target_attr] == positive_class]
data_neg = dataset[dataset[target_attr] != positive_class]
if len(data_pos) < len(data_neg):
train_pos, test_pos = split_dataset(data_pos, train_share)
train_neg, test_neg = split_dataset(data_neg, len(train_pos)/len(data_neg))
# set.difference makes the test set larger
test_neg = test_neg.iloc[0:len(test_pos)]
else:
train_neg, test_neg = split_dataset(data_neg, train_share)
train_pos, test_pos = split_dataset(data_pos, len(train_neg)/len(data_pos))
# set.difference makes the test set larger
test_pos = test_pos.iloc[0:len(test_neg)]
return train_pos.append(train_neg).sample(frac = 1).reset_index(drop = True), \
test_pos.append(test_neg).sample(frac = 1).reset_index(drop = True)
Usage:
train_ds, test_ds = split_dataset_stratified(data, target_attr, positive_class)
You can now perform cross validation on train_ds and evaluate your model in test_ds.

There is another solution that is in the model-level - using models that support weights of samples, such as Gradient Boosted Trees. Of those, CatBoost is usually the best as its training method leads to less leakage (as described in their article).
Example code:
from catboost import CatBoostClassifier
y = df['Label']
X = df.drop('Label',axis=1)
label_ratio = (y==1).sum() / (y==0).sum()
model = CatBoostClassifier(scale_pos_weight = label_ratio)
model.fit(X, y)
And so forth.
This works because Catboost treats each sample with a weight, so you can determine class weights in advance (scale_pos_weight).
This is better than downsampling, and is technically equal to oversampling (but requires less memory).
Also, a major part of treating imbalanced data, is making sure your metrics are weighted as well, or at least well-defined, as you might want equal performance (or skewed performance) on these metrics.
And if you want a more visual output than sklearn's classification_report, you can use one of the Deepchecks built-in checks (disclosure - I'm one of the maintainers):
from deepchecks.checks import PerformanceReport
from deepchecks import Dataset
PerformanceReport().run(Dataset(train_df, label='Label'), Dataset(test_df, label='Label'), model)

your implementation of stratified train/test creation is not optimal, as it lacks randomness. Very often data comes in batches, so it is not a good practice to take sequences of data as is, without shuffling.
as #sturgemeister mentioned, classes ratio 3:7 is not critical, so you should not worry too much of class imbalance. When you artificially change data balance in training you will need to compensate it by multiplication by prior for some algorithms.
as for your "perfect" results either your model overtrained or the model is indeed classifies the data perfectly. Use different train/test split to check this.
another point: your test set is only 94 data points. It is definitely not 1/5 of 1400. Check your numbers.
to get realistic estimates, you need lots of test data. This is the reason why you need to apply Cross Validation strategy.
as for general strategy for 5-fold CV I suggest following:
split your data to 5 folds with respect to labels (this is called stratified split and you can use StratifiedShuffleSplit function)
take 4 splits and train your model. If you want to use under/oversampling, modify the data in those 4 training splits.
apply the model to the remaining part. Do not under/over sample data in the test part. This way you get realistic performance estimate. Save the results.
repeat 2. and 3. for all test splits (totally 5 times obviously). Important: do not change parameters (e.g. tree depth) of the model when training - they should be the same for all splits.
now you have all your data points tested without being trained on them. This is the core idea of cross validation. Concatenate all the saved results, and estimate the performance .

Cross-validation or held-out set
First of all, you are not doing cross-validation. You are splitting your data in a train/validation/test set, which is good, and often sufficient when the number of training samples is large (say, >2e4). However, when the number of samples is small, which is your case, cross-validation becomes useful.
It is explained in depth in scikit-learn's documentation. You will start by taking out a test set from your data, as your create_cv function does. Then, you split the rest of the training data in e.g. 3 splits. Then, you do, for i in {1, 2, 3}: train on data j != i, evaluate on data i. The documentation explains it with prettier and colorful figures, you should have a look! It can be quite cumbersome to implement, but hopefully scikit does it out of the box.
As for the dataset being unbalanced, it is a very good idea to keep the same ratio of labels in each set. But again, you can let scikit handle it for you!
Purpose
Also, the purpose of cross-validation is to choose the right values for the hyper-parameters. You want the right amount of regularization, not too big (under-fitting) nor too small (over-fitting). If you're using a decision tree, the maximum depth (or the minimum number of samples per leaf) is the right metric to consider to estimate the regularization of your method.
Conclusion
Simply use GridSearchCV. You will have cross-validation and label balance done for you.
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=1/5, stratified=True)
tree = DecisionTreeClassifier()
parameters = {'min_samples_leaf': [1, 5, 10]}
clf = GridSearchCV(svc, parameters, cv=5) # Specifying cv does StratifiedShuffleSplit, see documentation
clf.fit(iris.data, iris.target)
sorted(clf.cv_results_.keys())
You can also replace the cv variable by a fancier shuffler, such as StratifiedGroupKFold (no intersection between groups).
I would also advise looking towards random trees, which are less interpretable but said to have better performances in practice.

Just wanted to add thresholding and cost sensitive learning to the list of possible approaches mentioned by the others. The former is well described here and consists in finding a new threshold for classifying positive vs negative classes (generally is 0.5 but it can be treated as an hyper parameter). The latter consists on weighting the classes to cope with their unbalancedness. This article was really useful to me to understand how to deal with unbalanced data sets. In it, you can find also cost sensitive learning with a specific explanation using decision tree as a model. Also all other approaches are really nicely reviewed including: Adaptive Synthetic Sampling, informed undersampling etc.

How can I solve a classification problem with a dependent variable with more than two values

I have a simple NLP problem, where I have some written reviews that have a simple binary positive or negative judgement. In this case I am able to train and test as independent variables the columns of X that contain the "bags of words", namely the single words in a sparse matrix.
from sklearn.feature_extraction.text import CountVectorizer
cv = CountVectorizer(max_features = 300)
#indipendent
X = cv.fit_transform(corpus).toarray()
#dependent
y = dataset.iloc[:, 1].values
..and the dependent variable y, that is represented by the column 1 that assume values as 0 and 1( so basically positive and negative review).
if instead of 0 and 1, I have reviews that can be voted from 1 to 5 stars should I proceed having an y variable column with values from 0 to 4?In other words I would lie to know how differ the model if instead of a binary good/bad review, the user has the possibility after his or her review to give a rating from 1 to 5.
How is called this kind of problem in machine learning?

It is just multi-class classification problem. Here is a sample code from where you can get an idea. What you are calling 'dependent variable' is called class (class that the input example belongs to)
label_idx = [unique.index(l) for l in labels] """ labels= class. works for your class is string or so.
here labels can be more than two"""
label_idx = np.array(label_idx) # just get your class into array
vectors = np.array(vecs) # vecs are any vectorised form of your text data
clf = LinearSVC() # classifier of your choice
clf.fit(vectors, label_idx)

I have used the following link for a RandomForest multiClassifier which is one of many possible ML algorithms you can use:
https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html#sklearn.ensemble.RandomForestClassifier
However, my personal experience shows deep learning neural networks work better with "text data" and tree-based models are better for tabular data with numeric values.

This problem is called as multi-class classification problem as mentioned by #rishi. There is a large variety of algorithms that can solve the multi-class problem. Look here
You could make your target variable as one, which as the ratings.
#dependent
y = dataset.iloc[:, 'ratings'].values
Then, you can fit this data into the classifier!
from sklearn import linear_model
clf = linear_model.SGDClassifier()
clf.fit(X, y)

Why can't I predict new data using SVM and KNN?

I'm new to machine learning and I just learned KNN and SVM with sklearn. How do I make a prediction for new data using SVM or KNN? I have tried both to make prediction. They make good prediction only when the data is already known. But when I try to predict new data, they give an incorrect prediction.
Here is my code:
import numpy as np
from sklearn import svm
x=np.array([[1],[2],[3],[4],[5],[6],[7],[8],[9],[10],[11]], dtype=np.float64)
y=np.array([2,3,4,5,6,7,8,9,10,11,12], dtype=np.float64)
clf = svm.SVC(kernel='linear')
clf.fit(x, y)
print(clf.predict([[20]]))
print(clf.score(x, y))
0utput:
[12.]
1.0
This code will make a good prediction as long as the data to predict is within the range x_train. But when I try to predict for example 20, or anything above the range x_train, the output will always be 12 which is the last element of y. I don't know what I do wrong in the code.

The code is behaving as mathematically described by a support vector machine.
You must understand how your data are being interpreted by the algorithm. You have 11 data points, and you are giving each one a different class. The SVM ends up basically dividing the number line into 11 segments (for the 11 classes you defined):
data = [(x, clf.predict([[x]])[0]) for x in np.linspace(1, 20, 300)]
plt.scatter([p[0] for p in data], [p[1] for p in data])
plt.show()
The answer by AILearning tells you how to fit your given toy problem, but make sure you also understand why your code wasn't doing what you thought it was. For any finite set of examples there are infinitely many functions that fit the data. Your fundamental issue is you are confusing regression and classification. From the sounds of it, you want a simple regression model to extrapolate a fit function from the data points, but your code is for a classification model.

You have to use a regression model rather than a classification model. For svm based regression use svm.SVR()
import numpy as np
from sklearn import svm
x=np.array([[1],[2],[3],[4],[5],[6],[7],[8],[9],[10],[11]], dtype=np.float64)
y=np.array([2,3,4,5,6,7,8,9,10,11,12], dtype=np.float64)
clf = svm.SVR(kernel='linear')
clf.fit(x, y)
print(clf.predict([[50]]))
print(clf.score(x, y))
output:
[50.12]
0.9996

Naive Bayes probability always 1

I started using sklearn.naive_bayes.GaussianNB for text classification, and have been getting fine initial results. I want to use the probability returned by the classifier as a measure of confidence, but the predict_proba() method always returns "1.0" for the chosen class, and "0.0" for all the rest.
I know (from here) that "...the probability outputs from predict_proba are not to be taken too seriously", but to that extent?!
The classifier can mistake finance-investing or chords-strings, but the predict_proba() output shows no sign of hesitation...
A little about the context:
- I've been using sklearn.feature_extraction.text.TfidfVectorizer for feature extraction, without, for start, restricting the vocabulary with stop_words, or min/max_df --> I have been getting very large vectors.
- I've been training the classifier on an hierarchical category tree (shallow: not more than 3 layers deep) with 7 texts (manually categorized) per category. It is, for now, flat training: I am not taking the hierarchy into account.
The resulting GaussianNB object is very big (~300MB), and prediction is rather slow: around 1 second for one text.
Can this be related? Are the huge vectors at the root of all this?
How do I get meaningful predictions? Do I need to use a different classifier?
Here's the code I'm using:
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import GaussianNB
import numpy as np
from sklearn.externals import joblib
Vectorizer = TfidfVectorizer(input = 'content')
vecs = Vectorizer.fit_transform(TextsList) # ~2000 strings
joblib.dump(Vectorizer, 'Vectorizer.pkl')
gnb = GaussianNB()
Y = np.array(TargetList) # ~2000 categories
gnb.fit(vecs.toarray(), Y)
joblib.dump(gnb, 'Classifier.pkl')
...
#In a different function:
Vectorizer = joblib.load('Vectorizer.pkl')
Classifier = joblib.load('Classifier.pkl')
InputList = [Text] # One string
Vec = Vectorizer.transform(InputList)
Probs = Classifier.predict_proba([Vec.toarray()[0]])[0]
MaxProb = max(Probs)
MaxProbIndex = np.where(Probs==MaxProb)[0][0]
Category = Classifier.classes_[MaxProbIndex]
result = (Category, MaxProb)
Update:
Following the advice below, I tried MultinomialNB & LogisticRegression. They both return varying probabilities, and are better in any way for my task: much more accurate classification, smaller objects in memory & much better speed (MultinomialNB is lightning fast!).
I now have a new problem: the returned probabilities are very small - typically in the range 0.004-0.012. This is for the predicted/winning category (and the classification is is accurate).

"...the probability outputs from predict_proba are not to be taken too seriously"
I'm the guy who wrote that. The point is that naive Bayes tends to predict probabilities that are almost always either very close to zero or very close to one; exactly the behavior you observe. Logistic regression (sklearn.linear_model.LogisticRegression or sklearn.linear_model.SGDClassifier(loss="log")) produces more realistic probabilities.
The resulting GaussianNB object is very big (~300MB), and prediction is rather slow: around 1 second for one text.
That's because GaussianNB is a non-linear model and does not support sparse matrices (which you found out already, since you're using toarray). Use MultinomialNB, BernoulliNB or logistic regression, which are much faster at predict time and also smaller. Their assumptions wrt. the input are also more realistic for term features. GaussianNB is really not a good estimator for text classification.