Label Logic Modification (Runtime Engine)

[vkatari10]

ML Signal Generation

• Take relationship values and multiply them against the weights
• Label should be an aggregated sum of all those relationship values multiplied by the weights
• But the DF should fit them to where its only 1, 0, -1
• We can still access the raw final sig value by recalculating
• And then feed this to the risk engine to determine strength of signal instead of 1, 0, -1

Go Runtime Engine

• Modify the relationship updater to account for these changes