Train AutoML Model
Last updated
Last updated
Automated machine learning (AutoML) is the process of applying machine learning (ML) models to real-world problems using automation. More specifically, it automates the selection, composition and parameterization of machine learning models.
In Fluid AI - Frontier Analytics Platform we have integrated many open-source AutoML tools as well as added traditional machine learning algorithms with automation in the pipeline.
To run using AutoML, In the Data Tab under Input File or Output File section select the DataFile, Click on , ▶ Preview File Then,
Click on Train AutoML Model.
Next you have to choose the Target Column from the dropdown once selected, Dashboard will automatically detect the type of problem e.g regression, binary-classification, multi-classification. It also give you privilege of choosing the problem type.
Primary key is the column which has unique-id for every row in the dataset. If you don't have any columns for that you can ignore it.
Also you can choose which AutoML template to select, SimpleAutoMl is for quick and fast prototyping the ML models and AdvanceAutoML is for a developer who want more control on process
Selects the columns on which you want to train the model and click on Create ML Notebook, you will get pop-up of success once done click on Go to Notebook
Page will redirect you to notebook/kernel.
You can run each cell either by shit + Enter, alternatively run all cel by a click on an icon ⏩ on top of taskbar.
In this template we have integrated EvaML opensource-tool. This use same api template as Sklearn for training and testing model. All the required preprocessing of Data such as Data-Cleaning, Imputation and Encoding for categorical data is being already integrated. Also text-vectorization for long string data will be automatically performed. You can save and load the model by using same template as Sklearn. Useful-link :
In this template we have integrated AutoML as well as traditional ML opensource library . Its automatically preprocess as well as vectorized the data. It trains data on the model such as AutoKeras, AutoSklearn, XGBOOST, RandomForest. You can also add different models in the pipeline as well. Useful-link :
In the pipeline section -> Build Pipelines -> Train AutoML Model, users can configure the advance Automl. Once click on Train Model, it will create the notebook and auto execute the all cells. Moreover, it will create new models which can be seen in the Model Registry section.