Custom Components

This repository contains a collection of machine-learning components for various tasks.

Components and Tasks

1. Vector Assembler

   • This component allows you to easily combine multiple feature columns into a single feature vector.

2. Split Data

   • This component allows you to specify the percentage of data allocated for training and testing.

3. Train Linear Regression

   • This component is designed to train your linear regression model.
   • You can specify various model parameters such as:
     • Target column
     • Maximum iterations
     • Regularization parameter
     • Elastic net mixing parameter
   • The trained model will be saved to a location specified by modelPath. For example: modelPath="dbfs:/FileStore/lr_model"

4. Predictions Linear Regression

   • This component is designed to make predictions based on loading a pre-trained model and applying it to new data for prediction.

5. Evaluations Linear Regression

   • This component is designed to evaluate the performance of the linear regression model by providing the Root Mean Squared Error (RMSE).

6. Load Data from Hugging Face

   • This component is designed to load datasets from the Hugging Face's datasets library and prepare data for Natural Language Processing (NLP).

7. Convert Word to Vector

   • This component is designed to generate word embeddings from text data.
   • You can configure word2vector parameters such as:
     • Input Column
     • Output Column
     • Vector Size
     • Minimum Count
     • Number of partitions
     • Step size
     • Maximum Iterations
     • Window Size
     • Max Sentence Length
     • Model path
   • Once word embeddings are created, you can save the trained model to a specified location for later use. For example: modelPath="dbfs:/FileStore/trainedModel"