delorian

Retrofit Expected Solana Transactions Clustering with Delta-Timing. This by building a mathematical expectation model, valid for a delta-in-time between blocks, capable of forecasting upcoming clusters of solana transactions with unusual jito tips.

Project's elements:

• delorian-base: General data structures and core types.
• delorian-data: Data acquisition, pre-processing and I/O engines.
• delorian-mods: Model Fitting/Inference engines.
• delorian-rs: re-exporting components into a single engine and docs.

Build Data-layer

• Specify credentials
• Define delivery method
  • default: minute batches in a .parquet file.
  • option 1: send to terminal.
  • option 2: write into a DB.
• Provision a TAE (Transaction-Acquisition-Engine)

Fit Model

• Define parameters for fitting.
• Load/Validate Data Set.
• Fit and report results.
• Save model's parameters.
• Save model's binary.

Serve Model

• use of dockerfile:
• load model's parameter and binary
• specify data input: Read from DB, include a TAE in the Serving.