[IGNITE-12079] [ML][Umbrella] Add advanced preprocessing techniques

• Type: New Feature
• Status: In Progress
• Priority: Major
• Resolution: Unresolved
• Component/s: ml
• Ignite Flags: Docs Required, Release Notes Required
• Assignee: Alexey Zinoviev
• Reporter: Alexey Zinoviev
• Created: 16/Aug/19 09:31
• Updated: 02/Dec/20 12:00

Description

Main goal:

To reduce the gap between Apache Spark and Apache Ignite in preprocessing operations. The reducing of the gap could help with loading Spark ML Pipelines to Ignite ML.

 

Next steps:

1. Add Frequency Encoder
2. Add two Imputing Strategies (MIN, MAX, COUNT, MOST_FREQUENT, LEAST_FREQUENT)
3. Add RobustScaler (will be added in Spark 3.0)
4. Add CountVectorizer
5. Add FeatureHasher
6. Add QuantileDiscretizer
7. Add Locality Sensitive Hashing (LSH)
8. Add LabelEncoder
9. Add RevertStringIndexing
10. Add multi-column preprocessor

Sub-Tasks

• [ML] Add Frequency Encoding: Resolved
• [ML] Add support of the additional Imputing Strategies: Resolved