This method selects partners falling within a user defined performance range for further analysis. The dealers are ordered by increasing forecasted performance. The horizontal axis represents the cumulative number of dealers and the vertical axis represents the expected change in dealer order input performance. Hovering over the curve identifies the dealer identity.
The sliders set the low and high performance range desired for further examination or analysis. This allows deeper inspection across all dealers based on expected performance.
Setting the Lower Partner slider to 400 and the Upper Partner slider at 500 reveals that only 60 dealers are going to perform with unexpected high performance in the next quarter. Allocating additional resources to these 60 potential high performers could assure the necessary success in the newly competitive environment.
Observing the area under the curve shows that about 200, or 40% of the dealers cumulatively, dealers are going to provides over 75% of the client's cumulative order input. This skewed cumulative performance argues that the client should apply more resources to this dealer group for maximum order input performance.
The low performing dealers do not deserve additional resources until further analysis yields the causes for their poor performance. Unless an effective plan to remedy the expected poor performance can be determined, allocating resources for low performance has a low return on its investment.
The hypothesis that a few dealers constitute the majority of high and low performers is confirmed. The opportunity presented to the client is to select these dealers and focus on the motivation for each separately. The next step will determine feature importances.Feature Importance Free Proof Of Concept