Ensuring that the data get into R correctly

[dill]

How can folks ensure that they loaded the gdb file correctly and that the data in it are correct?

[dill]

In the exercises I've used plot and head to try to show folks that the data is loaded correctly. My theory being if it's really messed-up it won't load. If it's a bit messed-up then they will be able to look at the plots in Arc and R and see if there's something weird happening.

Any other tips? (Don't really want to delve into calculating summary statistics etc.)

[jjrob]

plot is good; they will be used to seeing the data in Arc, so if it looks similar in a plot that is a good sign. It is also important to check the names, types, and values of the fields, to make sure that the Arc → R process did not mangle anything. Typical problems include:

• Fields were renamed (e.g. made shorter)
• Data types were poorly translated (e.g. dates turned into strings or numerics)
• NAs were not handled correctly (e.g. the arcgisbinding package turns NULLs in integer columns to -2147483647; this is probably a bug in that package)

head is good for that, but I often use summary instead, to look for min and max values and how many NAs there are. I also use class and typeof but this should probably not be necessary.

[dill]

I've added some more chat to the first practical about this in 653f2d7.

My advice:

• plot first
• use head to check column names and some values
• use summary to compare summary statistics
• use str to check data types -- this can be annoying for dates especially (here they are autoconverted to factor, but we don't care for now)

I've ignored the NA issue you raise as we're not dealing with that package. Can include if you think it's a major issue.