Recommendation Letter from Work Vs Another Professor

[gnelli]

I have been out of school for 5 years. I am applying to programs this fall and have 2 of my 3 recommendation letters lined up from Professors from undergrad. They are recognized in their respective fields and will be providing good reviews (they have disclosed them to me).

 

My question is, should I seek another professor for my 3rd recommendation? Or should this be a work reference? The work reference will provide a detailed recommendation of the quantitative work I've done in a Machine learning space, and applied statistics work (I work at a large software corporation).

 

If I were to seek another professor recommendation, it would either be a math professor with whom I have taken only 2 classes recently in preparation for Ph.D applications. I could try to find a coworker in my space with a Ph.D, but that connection may be a bit more loose.

 

Any/all opinions welcome. Much appreciated.

[tm_member]

I'd probably go with the math professor - but it depends on where the math professor works and their credentials.

[tbe]

I tend not to find letters of reference from employers/coworkers to be useful. While they always positive, they rarely talk about the skills necessary to do well in graduate school. So I think it is best to get recommendations from people who have a PhD, preferably in economics, but math is fine too as a third reference. You can talk about your background in machine learning and applied statistics in your own letter.

[gnelli]

I tend not to find letters of reference from employers/coworkers to be useful. While they always positive, they rarely talk about the skills necessary to do well in graduate school. So I think it is best to get recommendations from people who have a PhD, preferably in economics, but math is fine too as a third reference. You can talk about your background in machine learning and applied statistics in your own letter.

 

I appreciate this insight. My follow-up question would be, what if the letter spoke to both quantitative aspects of my work and my ability to lead research. Some examples could include:

- My ability to utilize contemporary machine learning models, e.g. models based on bayesian inference, or creating forecasts using appropriate ARIMA or ETS method in Python.

- Led research for my organization in these statistical methods and technologies. Created the organizations first set of models utilizing SAS, Python, MatLab, and other relevant technologies.

 

Appreciate initial input and any additional feedback.