Milestone 3 sanjosespartans TrustIdea 1: Fair Pricing Model

From crowdresearch
Jump to: navigation, search


There a a clear need for a system which can predict fair prices to the requesters so that:

• Requesters don’t have a hard time estimating the price of their task

• the requester’s don’t underpay or over pay

• a task is duely compensated


Regression models can be used to at first learn what fair prices for different tasks are and then predicting a price for a new task in hand giving the requesters an idea of what an ideal price for the task in the market look like.

Regression models have alrady been used fairly commonly in stock markets to predict market shares based on past information. Given enough relevant data (skills, past pricing, etc.) the model can be expected to learn a predictive function for guessing a reasonable price.


A fair price predictive model which would suggest what the price of a task should be in the market. Further restrictions can be enforced for the task price to not drop beyond certain threshold (say 30% of suggested price) to ensure a task is reasonably priced.


When the requester is ready to price his/her task, suggest a fair price based on the regression model’s experience.

A fair pricing model can learn a function to predict a fair price based on the previous price tags on tasks. Important predictive features for judging a price can be based on the skill sets required for the task, time commitment required, worker’s experience, time limit put on the task, etc.

Based on the price predicted by the fair pricing model, a threshold can be set to prevent the price of the task to fall below that (e.g. 30 % of predicted value). This would help in ensuring that the task is not underrated.