estimating the value of a property (real estate) using the hedonic regression I'm trying to estimate the value of a property depending on the property characteristics. I did some research and I found out, that it would be better to use the Hedonic Model/Regression instead of Linear Square Regression.
After reading a couple of papers about it, I still have some questions.
I work with R, so I have the data (information about other properties) saved as a data.frame, with the following columns (c stands for characteristic).
----------------------------------
| price | c1 | c2 | c3 | c4 | c5 |
----------------------------------

My questions:


*

*I know how to estimate the coefficients with the Least Square Regression, but how do I do it with the Hedonic Regression? I know, that in R is no function for it.

*The environment characteristics (air pollution, criminality rate, etc.) are almost the same, because the properties are in the same district. The Last Square Regression gives them a very small coefficient, but they have a big importance in real life. How can I tell the regression, that they have a big importance?

*As I understood so far, if an attribute of an observation is missing, I should not use the observation, is that right?

*In the calculation of the coefficients, should I use only the date from nearby (example: same district) real estates or it would be better to use all real estates from the town?


Could somebody please give me a hint?
Thank you very much!       
 A: 1) There is no such thing as a hedonic regression as estimation method, you will use least squares / maximum likelihood estimator. 
2) I understood that your goal is to estimate effects of different characteristic, it might be that your understanding concerning different factors affecting pricing is not complete. Of course if you have insufficient variation in the factors then it might not be possible to estimate these effects. If you have only price information from the similar nearby districts then this might happen. 
3) Estimation methods typically eliminate incomplete records, you might try to imputate missing records for example by using characteristics from the nearby observations. 
4) Of course you should have enought variation in the characteristics and prices to isolate effects from the different factors. 
A: Some hints on your question 2 re environmental characteristics.
A. Care is needed in interpreting estimated coefficients as small, since their size will depend on the units in which both the independent and the dependent variables are measured.  It may be helpful to consider what the percentage effect on the dependent variable is of a 1% change in an independent variable.
B. Characteristics such as air pollution may be subject to threshold effects, that is, they only become important when they exceed a certain level.  To allow for this, you might consider using as an independent variable not the absolute level but the excess over a threshold level (and zero if below that level).  They may also vary over time, with occasional peak levels being much more important than average levels.  A suitable variable then might be 'number of days in a year on which a threshold level was exceeded'. Having said that, you do as Analyst has said need sufficient variation in the characteristics and prices to obtain meaningful results.
