Train, Test, Validate split Python. Three sets - python

Someone presented a solution to split a dataset into three sets. I wonder where is the label in this case. Or how to set the labels then.
train, validate, test = np.split(df.sample(frac=1), [int(.6*len(df)), int(.8*len(df))])

I will answer the question based on comments:
Using this method for splitting:
train, validate, test = np.split(df.sample(frac=1), [int(.6*len(df)), int(.8*len(df))])
You are getting 3 different objects, which consist of the first 60% of data from df for train, the data corresponding to the interval between 60% and 80% for validate and the last 20% corresponding to 80%-100% in test. The labels are within these dataframes.
In train_test_split you are passing two objects, X and Y, which have been most likely previously split from an original dataset and getting in return 4 objects, 2 corresponding to train and two corresponding to test. Keep in mind this: You are first splitting your dataset into independent variables and explained/target variable, and then splitting these two objects into train and test.
With np.split you are going the otherway around, you are first splitting your dataset into 3 objects, train, validate and test which will later need to be split individually into independent variables commonly known as X and target variable known as Y. You are doing the same splits, just in reverse order.
However, keep in mind that by passing the indexes for np.split it means the splitting is not performed randomly, whereas with train_test_split you get a random train and test subesets. np.split on the other hand, allows for more flexibility, for instance, as you prove with your example, creating more than 2 subsets.
Maybe this will help!

Try this. Feed the output of one of the train_test_split into a second one as input
import numpy as np
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.4)
X_test, X_validate, y_test, y_validate = train_test_split(X_test, y_test, test_size=0.5)
The function randomly splits 2 arrays into 4 arrays, and test_size determines the size of the split allocated to the test output vs train. The y input is meant to be a target for building a machine learning model and X is meant to be the features for the model. If you want them combined, then just concat the equivalent X and y outputs.

Related

Pairwise comparisons for model training/testing - how to parameter tune?

For some reasons, I have base dataframes of the following structure
print(df1.shape)
display(df1.head())
print(df2.shape)
display(df2.head())
Where the top dataframe is my features set and my bottom is the output set. To turn this into a problem that is amenable to data modeling I first do:
x_train, x_test, y_train, y_test = train_test_split(df1, df2, train_size = 0.8)
I then have a split for 80% training and 20% testing.
Since the output set (df2; y_test/y_train) is individual measurements with no inherent meaning on their own, I calculate pairwise distances between the labels to generate a single output value denoting the pairwise distances between observations using (the distances are computed after z-scoring; the z-scoring code isn't described here but it is done):
y_train = pdist(y_train, 'euclidean')
y_test = pdist(y_test, 'euclidean')
Similarly I then apply this strategy to the features set to generate pairwise distances between individual observations of each of the instances of each feature.
def feature_distances(input_vector):
modified_vector = np.array(input_vector).reshape(-1,1)
vector_distances = pdist(modified_vector, 'euclidean')
vector_distances = pd.Series(vector_distances)
return vector_distances
x_train = x_train.apply(feature_distances, axis = 0)
x_test = x_test.apply(feature_distances, axis = 0)
I then proceed to train & test all of my models.
For now I am trying linear regression , random forest, xgboost.
Is there any easy way to implement a cross validation scheme in my dataset?
Since my problem requires calculating pairwise distances between observations, I am struggling to identify an easy way to do cross validation schemes to optimize parameter tuning.
GridsearchCV doesn't quite work here since in each instance of the test/train split, distances have to be recomputed to avoid contamination of test with train.
Hope it's clear!
First, what I understood from the shape of your data frames that you have 42 samples and 1643 features in the input, and each output vector consists of 392 values.
Huge Input: In case, you are sure that your problem has 1643 features, you might need to use PCA to reduce the dimensionality instead of pairwise distance. You should collect more samples instead of 42 samples to avoid overfitting because it is not enough data to train and test your model.
Huge Output: you could use sampled_softmax_loss to speed up the training process as mentioned in TensorFlow documentation . You could also read this here. In case, you do not want to follow this approach, you can continue training with this output but it takes some time.
x_train, x_test, y_train, y_test = train_test_split(X, y, test_size=0.8, random_state=n)
here X is independent feature, y is dependent feature means what you actually want to predict - it could be label or continuous value. We used train_test_split on train dataset and we are using (x_train, y_train) to train model and (x_test, y_test) to test model to ensure performance of model on unknown data(x_test, y_test). In your case you have given y as df2 which is wrong just figure out your target feature and give it as y and there is no need to split test data.

xgboost - feature mismatch when I predict on my test data

I"m using xgboost to train some data and then I want to score it on a test set.
My data is a combination of categorical and numeric variables, so I used pd.get_dummies to dummy all my categorical variables. training is fine, but the problem happens when I score the model on the test set.
I get an error of "feature_names_mismatch" and it lists the columns that are missing. My dataset is already in a matrix (numpy array format).
the mismatch in feature name is valid since some dummy-categories may not be present in the test set. So if this happens, is there a way for the model to still work?
If I understood your problem correctly; you have some categorical values which appears in train set but not in test set. This usually happens when you create dummy variables (converting categorical features using one hot coding etc) separately for train and test instead of doing it on entire dataset. Following code can help
for col in featurs_object:
X[col]=pd.Categorical(X[col],categories=df[col].dropna().unique())
X_col = pd.get_dummies(X[col])
X = X.drop(col,axis=1)
X_col.columns = X_col.columns.tolist()
frames = [X_col, X]
X = pd.concat(frames,axis=1)
X = pd.concat([X,df_continous],axis=1)
X_train, X_test, y_train, y_test = train_test_split(X, y,
test_size = 0.3,
random_state = 1)
featurs_object : consists of all categorical columns which you want to include for model building.
df : your entire dataset (post cleanup)
df_continous : Subset of df, with only continuous features.

Splitting test/training data for scikit?

I was given some starter code, but I'm not sure how to split it up when calling train_test_split (which I was explicitly told to use). Essentially, where does it come into play when I'm already given an X_train, Y_train, and X_test split?
The starter code looks like so:
train_df = pd.read_csv('./train_preprocessed.csv')
test_df = pd.read_csv('./test_preprocessed.csv')
X_train = train_df.drop("Survived",axis=1)
Y_train = train_df["Survived"]
X_test = test_df.drop("PassengerId",axis=1).copy()
print(train_df[train_df.isnull().any(axis=1)])
##SVM
svc = SVC()
svc.fit(X_train, Y_train)
Y_pred = svc.predict(X_test)
acc_svc = round(svc.score(X_train, Y_train) * 100, 2)
print("svm accuracy is:", acc_svc)
I need to change the acc_svc variable to be using X_test and Y_test, however. X_test is given to us, but how do I come up with a Y_test? I know the Y_test should correspond to labels, and I'm having some size mismatching going on when I attempt to do so. Should be a simple question, anyone mind pointing me in the right direction?
The test_preprocessed.csv shouldn't be used to check your model performance. Split your train_df using train_test_split() in scikit-learn into train and validation datasets. You have to check your model performance on validation dataset i.e. y of validation. Please refer to: scikit-learn documentation
First of all, you have to understand and clarify your target variable. Your "Y_test" seems to be your already existed "Y_pred" variable, which seems to correspond to the "Survived" label (in your test set). However, although you are dropping it from the "X_train" so that you can use it as a target, you don't seem to do the same in the "Y_train", where instead you are dropping "PassengerId".
Another basic concept here is that your dataset is already split into train-test subsets (your CSV files). I assume that your test set has already one less column compared to the train set, and that should be the "Survived" variable as a continuation from the train CSV file. Otherwise, you should drop it to avoid mismatching and keep that as your test target variable. You don't have to come up with a "Y_test", the result from your equation "Y_pred = svc.predict(X_test)" will give you the "Y_test" which would be the result of the "Y_pred".
One possible reason you get size mismatching is that the number of columns (x-axis) in your train set is not equal with that of the test set.
If you want to split into train/test subsets based on Scikit-learn you would first merge your CSV files, then do the data analysis in the merged dataset, and finally, do the split. One way to keep track of these changes and maintain the same original size of the train-test split could be to keep key-value pairs originated from the train-test merge. One way to do that could be via the pandas.concat, using the parameter "keys".
Incorporating the above, one recommended simple solution might be:
# reading csv files
train_df = pd.read_csv('./train_preprocessed.csv')
test_df = pd.read_csv('./test_preprocessed.csv')
# merge train and test sets
merged_data = pd.concat([train_df, test_df], keys=[0,1])
# data preprocessing can take place in the below assigned variable
# here also you could do feature engineering etc.
# e.g. check null values for all dataset
print(merged_data[merged_data.isnull().any(axis=1)])
# now you can eject the train and test sets, using the key-value pairs from the train-test merge
X_train = merged_data.xs(0)
X_test = merged_data.xs(1)
# setting up predictors - target
X= X_train.loc[:, X_train.columns!="Survived"]
y= X_train.loc[:, "Survived"]
# train-test split
# If train_size is None, it will be set to 0.25 based on the documentation
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=42)
##SVM
svc = SVC()
svc.fit(X_train, y_train)
y_pred = svc.predict(X_test)
acc_svc = round(svc.score(X_train, y_train) * 100, 2)
print("svm accuracy is:", acc_svc)
In my opinion, after understanding the above you could further estimate and compare your model's performance using the cross_val_score function, in a way #SunilG mentions. For e.g. a 3-fold (CV=3) cross validation, you could:
from sklearn.model_selection import cross_val_score
cross_val_score(svc, X_train, y_train.values, cv=3, scoring='accuracy')
If you do not want to proceed to the above and you want to be close to your starter code, then you should delete your 5th line of code and I suppose it would run (if your test set does not include your target variable, otherwise drop it). However in this case you would not be able to split your train-test on your own, since it is already split, hence the title of your main question/post should be altered.

Is train_test_split necessary for binary classification? And why are there 4 outcomes?

Why are there 4 outcomes to train_test_split in sklearn? Why is there y_test, if the testing data has no y_data?
The reason you get 4 outcomes is because you get: train_features, test_features, train_labels and test_labels (X_train, X_test, y_train, y_test). So it not just splits the dataset into train and test set, but also the labels. (so 2 + 2 = 4 outcomes).
Looking into the documentation, you can see that the first parameter is
*arrays, which means you can put as many arrays as you want there. Now, what does it returns?
Returns: splitting : list, length=2 * len(arrays)
Which means it returns twice the amount of arrays passed in the train_test_split function.
So, if you already have a training and a testing set, it only makes sense to split the training set, so you can have a validation set to check the model performance.
Eg.:
train_data, validation_data, train_label, validation_label= train_test_split(original_train_data, original_train_label)
Note that you also must split the labels in the case you have the data and the label in separated vectors.
because you have split your original data into train and test parts. so there would be four outcomes.
1 (X_train, Y_train) where X_train are the training points while Y_train are their respective class labels. Now this is your training data which will be used to train your model with any classical models like K-NN, logistic regression , Decision Tress.
2 (X_test,Y_test) where X_test represents your test data point and y_train are your respective class labels for these test points.Now once you have trained your model and calculated your training error/accuracy, then you can use these points to see whether the trained model predicts the data correctly or not.The lower the difference between your training and test error the better it is.
That is why you get 4 outcomes with pairs of 2 each.
Hope this helps.

what is the differene between Stratify and StratifiedKFold in python scikit learn?

My data consists of 99% target variable = 1, and 1% target variable = '0'. Does stratify guarantee that the train tests and test sets have equal ratio of data in terms of target variable. As in containts, equal amounts of '1' and '0'?
Please see below code for clarification
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2,stratify=y,random_state=42)
Stratification will just return a portion of data which may be shuffled or not based on the arguments you pass to it. let's say your dataset consists of 100 instances of class 1 and 10 instances of class 0, you decide to do a split of 70:30, suppose you pass the appropriate parameters to get a split of 63-class1 instances and 7-class0 instances in training set and 27-class1 instances and 3-class0 instances in the test set. Clearly, it is no way balanced. The classifier you train will be highly biased and as good as a dummy classifier which predicts every input as class1.
A better approach would be, either try to collect more data of class-0, or oversample the dataset to artificially generate more class0 instances or undersample it to get less class1 instances. python imblearn is a library in python which can help you for that
First difference is that the train_test_split(X, y, test_size=0.2, stratify=y) will only split the data once and in which 80% will be in train and 20% in test.
Whereas StratifiedKFold(n_splits=2) will split the data into 50% train and 50% test.
Second is that you can specify n_splits greater than 2 to achieve a cross-validation fold effect, in which the data will splitted n_split number of times. So there will be multiple divisions of data into train and test.
For more information about the K-fold you can look at this question:
difference between StratifiedKFold and StratifiedShuffleSplit in sklearn
The idea is same in that. train_test_split will internally use StratifiedShuffleSplit

Categories