Skip to content

Semi-supervised learning which apply a data science pipeline toward a multilayer perceptron binary classifier of loan status from Kaggle.

License

Notifications You must be signed in to change notification settings

yanncauchepin/Kaggle_LendingClub

Repository files navigation

Semi-Supervised Learning with Binary MLP Classifier

Here is a repository related to the dataset Lending Club.

Associated notebook is also pusblished here.

Docker Image

You can access to my trained MLP binary classifier with the docker image.

docker pull yanncauchepin/kaggle_lendingclub_mlp_binary_classifier:latest

How it works?

You can predict the loan_status of a new data by using this model from two data insertion: manually or from json file. It depends on the choice you make in the menu of the docker image. To access to the menu, you must run the docker image in interactive mode -it.

docker run -it yanncauchepin/kaggle_lendingclub_mlp_binary_classifier:latest

If you would like to use a json data, please ensure you provide data in the same format as the docker/data/data_test.json file and well write the feature names. To bring the json data file in the docker image you must add a volume by running the following command. Assuming you have a folder named data_folder which contains your json data file data.json, enter the following command. Please adapt the root_path information according to your system.

docker run -it -v /root_path/data_folder:/app/data yanncauchepin/kaggle_lendingclub_mlp_binary_classifier:latest

By choosing this second choice of predicting the loan_status from a data json file, you can get the output by passing data_folder/data.json as input into the request. Nevertheless, if you inform datat_test.json, or simply press Enter, it will run the inner test.


About

Semi-supervised learning which apply a data science pipeline toward a multilayer perceptron binary classifier of loan status from Kaggle.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages