New to the field of stats, trying to apply it to my day job (cost estimating) So to preface, I am very new to the world of Stats / quantitative analysis (I have recently gone back to uni and the paper I just completed is QA, specifically Regression, Estimation, Hyp. testing, Optimization.
I have really enjoyed the course but sadly never got a chance to apply it to my day job during the semester.
I work in the construction industry running the estimating department (i am a  tradesman by nature and have gained alot of experience at the coal face).
What I would ideally like is a regression model that can give me guidance on what the estimated margin should be according to historical records.  Do you think this is even achievable? Or is there anything you can see that would be possible from the info?
Another inclusion I considered was the actual end of job report when all costs and final sell price is known.
EM, Estimated Margin,   Dependent Variable, Best guess estimated margin 
E,  Estimator,      Independent,    Non-metric, Name
PH, Project hours,  Independent,    Man hours total
MT, Materials total,    Independent,    Dollar value of materials
WT, Work Type,      Independent,    Non-metric,technical,maintenance,Construction
S,  Successful,     Independent,    Yes, No, Pending, Cancelled (I can filter to only look at yes/no
I have loads of info but not sure where to start really start, I have tried running one model but excel crashed over and over again. (too much data for excel I guess, I do have SPSS but not 100% sure how to use it)
Currently the way we take a guess at where the margin should be, My chief estimator has 40+ years experience and he is always on the money... However he has 40+ years of experience and is looking to make a move to retirement.........
Any feedback or help would be greatly appreciated.
 A: To answer your broadest question, yes, it can be done. However, your results might have extremely large (and unacceptable) uncertainties depending upon your data and specific questions. If I were in your shoes, here are the broad steps I would take:


*

*Think about what questions you want to answer. You've started doing this, but try to be more specific. Start broad and then focus your question down (e.g., you are broadly interested in estimating margin the margin. What does this mean financially? What specifics do you have data for to help you understand this?). Ultimately, you'll end up with a quantitative relationship you can express algebraically, e.g, $ margin \sim predictors$.

*Use your quantitative data to choose and formulate a model. Start simple and and make the model more complex as needed. Make sure the model you choose makes sense and is reasonable given both the data and your question. Ask yourself questions about the model (e.g., do you meet the assumptions? does the model make sense?). 

*Choose the tools you need to solve your problem. Personally, I like R, but it has a steep learning curve if you're new to programing and it took me years to become good at using it. (@Mox's answer provides some useful tips for using R). Perhaps SPSS or Excel meet your needs. I would suggest the book Modeling for Insight because provides great tips for how to quantitatively think about problems in business. The book also provides a good tutorial for using spreadsheets for modeling.
In summary, think about your questions first and the tools second (and also ask your chief estimator how he thinks though his job!). Good luck!
A: First, get the R language for statistical computing and RStudio to the the analysis. 
Second, save it out of Excel as a csv, and import it into R, like so:
yourfile = read.csv("C:/directory/subdirectory/csvname.csv", header=TRUE)

Third, start running regression models. Basic syntax is like this:
Regressionmodel_1<- lm(EM ~ E +PH _MT+WT+S,data=yourdata) 

Fourth, start googling for help on R--it's plentiful, and you can copy-paste it in.
