How to do hyperparameter tuning with Dask?
This recipe helps you do hyperparameter tuning with Dask
Recipe Objective
How to do hyperparameter tuning with Dask
The hyperparameters controls the quality of the predictions made by models. For example, if learning rate is too small the model will take too much time to learn.
#!pip install dask_ml
#!pip install dask distributed --upgrade
Learn About the Application of ARCH and GARCH models in Real-World
Step 1- Importing Libraries
We will have to import multiple libraries to do the tuning of our hyper parameters
from dask.distributed import Client
from dask_ml.datasets import make_classification
from dask_ml.model_selection import train_test_split
from sklearn.linear_model import SGDClassifier
from dask_ml.model_selection import HyperbandSearchCV
from scipy.stats import uniform, loguniform
Step 2- Instantiating Client.
We will create the client and divide the dataset into train and test.
client = Client()
x, y = make_classification(chunks=20, random_state=0)
xtrain, xtest, ytrain, ytest = train_test_split(x, y)
Step 3- Instantiating SGD classifier.
We will instantiate SGDClassifier with its parameters.
cls = SGDClassifier(tol=1e-3, penalty='elasticnet', random_state=0)
Step 4- Initializing loss parameters.
We will define the learning rate and L1 loss.
parameters = {'alpha': loguniform(1e-2, 1e0), 'l1_ratio': uniform(0, 1)}
Step 5- Initializing HyperbandSearchCV
We will use the HyperbandSearchCV and insert classifier and parameter model in it to get the best parameters for the model.
search = HyperbandSearchCV(cls, parameters, max_iter=81, random_state=0)
search.fit(xtrain, ytrain, classes=[0, 1])
print(search.best_params_)
print(search.best_score_)
print(search.score(xtest, ytest))
What Users are saying..
Gautam Vermani
Having worked in the field of Data Science, I wanted to explore how I can implement projects in other domains, So I thought of connecting with ProjectPro. A project that helped me absorb this topic... Read More