The second step in creating on-chain scatter plots is regenerating code and PDFs from the code and data stored on the chain, facilitating the ability to easily compare algorithms and data sets.