Stat Computing > Limdep > FAQ
|
|
|
||||
|
Stat Computing > Limdep > FAQ
|
|
||||
The maximal number of variables that LimDep can handle is 199, not including the constant variable that LimDep uses for analysis purposes. If you have a data file that has more than 199 variables, you need to trim your data file down to less than 199 variables before the data file can be read in. Furthermore, you have to leave some room for new variables if you intend to create new variables. For example, assume that you have a categorical variable with ten levels and you have to create nine dummy variables for it. If you want to code the dummy variables in LimDep, you can only input at most 190 variables.
The easiest way to do the trimming is via Stat/Transfer where you can use the Variables Menu to select the variables that you want to keep. It also converts the data file to LimDep for Windows format at the same time. Stat/Transfer handles many different types of data file, including ASCII (e.g. tab delimited) files.
Without Stat/Transfer, you have to go back to the program where your data file was created and recreate a smaller data file with only the variables that you need for analysis in Limdep.
Within a session of LimDep, LimDep uses a fixed amount of memory. If you need to increase the memory that LimDep uses within a session, you are going to lose all the data inside the memory. Therefore, it is a good practice to carefully set up enough memory before inputting data and doing analyses. LimDep for Windows is set up with a default data area with 200,000 cells. This means that if a data file has 100 variables, then it can only have at most 2000 observations, not to say we may have to create new variables later, which means we need more cells. One way of allocating more memory for LimDep is through option Options in pull-down menu Tools.
Let's say we have a data file with 20 variables and 50000 observations. We need at least 20*50000=1,000,000 cells. Let's say we also want to create a few new variables. For each variable we create, we need 50000 cells. So if we need to create 10 more variables, we need 1,500,000 cells in total. Here is what we have to do to set up the memory for Limdep.
By now, LimDep is going to allocate enough memory up to 1,500,000 cells every time it starts.
- Start LimDep for Windows;
- Immediately go to the pull-down menu Tools and click on the choice Options;
Click on Projects;
Increase the number of Data Cells to what you need (in our case, to 1,500,000).
Click OK and QUIT from LimDep entirely. You don't have to save anything.
If for some reason you failed to input your data file, you can check the trace file to debug the problem. For example, we have a Stata data file with 100 variables and 20,000 observations. We converted it into LimDep for Windows using Stat/Transfer. When we tried to open the converted file in the default setting of LimDep, LimDep gave us an error message. It also suggested us to look at the trace file in the output window. The trace file (shown below) tells us the number of cells that this file has been set up to. We can accordingly allocate memory for Limdep and reenter our data file again.
--> LOAD;file="E:\limdep\data\large.lpj"$ +===========================================================+ |The data area settings embedded in this project file are | |larger than the settings currently in place for running | |LIMDEP/NLOGIT. Your current setting is 200000 cells. The | |project you are reading has a setting of 2200000 cells. | |This is a system setting that you must adjust before it is | |possible to continue. To do so, do the following steps: | |The Tools dialog box is in the desktop menu at the top | |left of your screen. It is the third to rightmost item. | |1. Go to Tools:Options/Projects tab and change the default | | memory allocation from 200000 to 2200000 or higher. | |2. Exit LIMDEP again, so that it can install the new | | system setting for the data area | |3. Restart LIMDEP. You can now reload this project file. | +===========================================================+
UCLA Researchers are invited to our Statistical Consulting Services
We recommend others to our list of Other Resources for Statistical Computing Help
These pages are Copyrighted (c) by UCLA Academic Technology Services
