Looking to hire freelance statistician

#1
I'm sorry if this isn't the right thread to post this question, but here goes.

I'm looking for a statistician to do some freelance work, but I'm not sure of the best place to find the right people. The work would involve some analysis of customer databases including developing some R scripts that could be used to "automate" some existing processes.

If you have any suggestions of where to look, please let me know.

Thanks.
JMR
 
#3
Fair question. I know exactly what I need in terms of scope, so I'd like to get a sense of how many hours would be involved and I'm more than happy to pay whatever prevailing hourly rate the right person uses (within reason). Once I get a sense of hours involved (and I know it won't be a tiny amount) and hourly rate, I can figure out if it makes sense. From work I've had done in the past, I don't think what I can pay will be an issue (unfortunately, people I've used in the past are no longer available).
 
#5
I'd estimate that it's probably 50-75 hours or so, but I could be way off. I'm open to paying rates in the $75-125 per hour depending on the experience of the person (and total project hours needed).
 
#6
So you talk about one or two weeks work (it could be more), and more than 100 US$ per hour.

Now, you need to be more specific about the work. Is it primarily a statistical consultation job (or is that already decided?) and a small part of it R programming?

Or is is it primarily, say 90%, an R programming work? Is it mostly cleaning of your data, or is it already tidy data? What kind of data input do you have and how do you want the results to look like? What does it mean ' to "automate" some existing processes.'?

From what country and in what language is the context of the data?
 
#7
Okay...here are some more details.

I have large databases of data consisting of clients' databases. Within the database are rows for each customer. Each customer then has 100-250 columns of data about them. This includes things like age, sex, income and things like interests (which are all Yes or No). So, it's a combination of categorical data and binomial data. What I'd like to do is analyze the data to find any natural clusters of people within the database taking into account all these different columns (attributes). Are there 4 or 5 clusters that naturally form, for example, and what attributes define each cluster? For each individual database (each different client), I would need to run the same analysis (the data structure is exactly the same for each client) again. So, I would need someone to set up that initial program to run the clustering (no idea of hierarchical, 2-way or something else makes sense) and then deliver an R script that I could run myself for subsequent databases (with the same structure).

Assume the data is already cleaned and normalized, as that's one of the first steps we handle on our end. The data are in English from the US.

Does that give enough detail?
 
#8
So you have a database that can be reached by say sql with demographics and characteristics of interests (a market research data set?) and an few other databases with essentially the same structure.

The clustering method has not been decided, but that is left to the statistician to decide.

What are the criteria for "success", having delivered? (Sometimes (here) someone suggest to pay when they or their boss is satisfied, in some remote time in the future, in a remote currency. That is very uncertain receivables. )

Does the OP Jonmrich have any knowledge herself in the coding of R? Or is it possible for Jonmrich to do small adaptations of the program?
 
#9
Not sure if the database will be in SQL or a csv, but in principle, you're correct with your first set of assumptions. Correct, the clustering method has not been decided. That's something I want to leave to an expert, but I expect to have the rationale explained to me, so that I can share the methods with clients if they ask. I'm the final decision maker on whether or not the results are acceptable. Assuming the clustering of the original data set seems reasonable and that the R script can be used successfully on another data set, I would declare the project complete and make payment.

I'm the OP and the one who would be working with the script after the fact. My knowledge of R coding is fairly limited, but I do have other coding experience, so I'd feel reasonably comfortable that I could do some small adaptations with a little help from some other contacts (or StackOverflow), but it would be good to have ongoing contact with the original developer of the script who I could consult with if something isn't working or I needed a bigger change (paid work, of course).