1
$\begingroup$

I have historical data for crowdsourcing micro tasks(Task : jobtype (classification of jobs(String)), location of job , price of the job, completion time(start date - final date)). These tasks are Human intelligence tasks such as in Amazon mechanical turk.

When a new micro crowdsourcing task comes up, I intend to compute optimum price for it. Is it possible to do through any machine learning algorithms, such as multiple regression model? If not any other suggestions are also appreciated.

If anyone could guide me in a direction, it would be really helpful.

Improve this question
$\endgroup$
1
  • $\begingroup$ Welcome to Cross Validated. It would be helpful if you can provide more information about the outcome variable. You mentioned the Cox PH model, so it seems you have information on survival/failure in your data, yet your question asked about "optimum price of a new given task." I'm having a hard time visualising how these two are related, especially if the task is categorical. What is meant by "new given task?" Please edit your question to elaborate and include some rows of your data. A screengrab would be sufficient $\endgroup$
    – Marquis de Carabas
    Commented Jul 11, 2016 at 6:36

1 Answer 1

Reset to default
2
$\begingroup$

Yes you can build a regression model to predict the 'Reward' based on given variables. You can't use cox for this but all other algos like multiple linear regression, decision trees, SVMs will be fine.

But without any more details about the data I can not comment whether you really need one & which model will be suitable.

Improve this answer
$\endgroup$
2
  • $\begingroup$ hisotrical data has set of tasks , and each task has reward, type of job(description) , location of the task to be performed, completion time(start date - completed date),. There is other trivial details for instance if task is verified or not, but important preedictors are the one i mentioned. Do you suggest any particular algorithm or tool or for this purpose? For instance weka regression. $\endgroup$
    – asaini
    Commented Jul 11, 2016 at 7:05
  • $\begingroup$ Since you have categorical variables I would suggest start with Random Forest (more of a personal opinion). Ideally you should try few algorithms & based on performance choose the one which has best performance. $\endgroup$
    – wololo
    Commented Jul 11, 2016 at 7:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.