Kata

Hi there! Welcome to "Building an Adaboost model with Sklearn (Introductory Machine Learning)".

If you get stuck at any point I strongly recommend you read the relevant documentation: http://scikit-learn.org/stable/index.html

This kata can be broken up into the following steps:

1. Import the relevant sklearn libraries (i.e. AdaBoostClassifier, train_test_split)
2. Create the "train_ada_boost" function (more details below)
3. Split the data (X) into a test set and a training set (you MUST USE sklearns train_test_split for this)
4. TRAIN an adaBoostClassfier (Once again, you MUST USE sklearn for this).
5. Your "train_ada_boost" function must return a three element tuple (more details below).

The "train_ada_boost" function accepts the following arguments:

• X -- This would normally be a dataset (but in this kata it is a 1D numpy array)
• Target -- This is a 1D numpy array consisting of 1's and 0's. This argument is the set of values we are trying to predict with our model
• estimators -- KEYWORD ARGUMENT if no argument is passed in the default value should be set to 3.
• random_seed -- KEYWORD ARGUMENT if no argument is passed in the default value should be set to 0.

Your function should return a 3 element tuple consisting of the following values in this exact order:

• A TRAINED AdaBoostClassifer (return the actual model)
• A test set (1D numpy array, which was built by sklearns test_train_split function)
• A target array (1D numpy array, which was built by sklearns test_train_split function))

Details:

• Your model should be trained using the specified number of estimators, with a random state equal to seed.
• When splitting the data, be sure to set the random state equal to seed.
• ALL other parameters not mentioned here for both 'test_train_split' and 'AdaBoost' should be set to default values.
• Even though you are handling Numpy arrays there is no need to actually manipulate the arrays yourself (let sklearn to it for you!).

Good Luck!

*Note, in actual machine learning it is highly unlikely we would ever use AdaBoost with such a small number of estimators. However building models can be very time consuming and codewars.com has a 5000ms timeout.