The capstone project class will allow students to create a usable/public data product that can be used to show your skills to potential employers. Projects will be drawn from real-world problems and will be conducted with industry, government, and academic partners.
Regular Expressions: Part 2 (Optional)
Loading...
来自 Johns Hopkins University 的课程
Data Science Capstone
699 个评分
从本节课中
Overview, Understanding the Problem, and Getting the Data
This week, we introduce the project so you can get a clear grip on the problem at hand and begin working with the dataset.
与讲师见面
Jeff Leek, PhDAssociate Professor, Biostatistics
Bloomberg School of Public Health
Roger D. Peng, PhDAssociate Professor, Biostatistics
Bloomberg School of Public Health
Brian Caffo, PhDProfessor, Biostatistics
Bloomberg School of Public Health
This is a continuation of our lecture on regular expressions.
So we're going to continue talking about Metacharacters,
which we can think of as the grammar where the literals are actually the words in
the language that we're going to be talking about.
So another Metacharacter is actually just the dot.
And so, the dot means it could be any character.
So if you search for the term 9.11, what that's going to look for is any
place where it sees a 9, followed by any possible character, followed by an 11.
So it matches all of these phrases because there's a 9 and
an 11 separated by exactly one character in all of these different examples.
Another metacharacter is the or metacharacter, and so what it can do,
be used is to combine two different expressions.
The subexpressions are called alternatives.
So here, it's combining the expression flood and the expression fire, and
they are the two alternatives.
So it will match all of these lines because there's either fire or
flood in every single line that we see here.
You can include up to any number of alternatives that you'd like.
So here it can match flood, earthquake, hurricane or a coldfire.
So all of these different lines get matched,
even though it's a different character that is identif, or
a different literal that's identifying in every single one.
The alternatives can also be expressions, regular expressions and not just literals.
So, for example, what we're searching for here is the beginning of the line, and
then either a capital or lowercase good, or anywhere in the line of bad.
So, for example, here you see good at the beginning of the line, and
here you see bad.
And it doesn't have to necessarily be at the beginning of the line,
because the alternative here doesn't have a carrot at the beginning.
An alternative is to include parenthesis to make sure to constrain the,
the alternatives.
So here we're looking at the beginning of the line and you have to have either good
or bad lower case or up, or upper case at the beginning of the line.
But in this case we've now constrained, even if we're looking for
bad it has to find it at the beginning of the line.
If you have a question mark that follows a parenthesis like this,
it suggests that the indicated expression is optional.
So here we have a lowercase or capital W, and then we have a period.
And it's an optional, in this particular phrase.
So we're looking for George Bush, with potentially the W in the middle.
But it'll ma, match this line because it doesn't necessarily have to have the W
in the middle to be able to find it.
Something to note is that here we actually had to escape the character dot, and
that's because dot, it can be viewed as a metacharacter.
And so by adding this backslash, what we've done is said, don't consider this to
be a, a metacharacter, consider it to be the literal dot.
Some more metacharacters are star and the plus signs.
So star means repeat any number of times, including none of the item.
And plus means at least one of the item.
So here we were searching for something between parenthesis and
it can be any character repeated any number of times.
So that will match everything from something like this where you
have a phrase in between the parenthesis, to something like this where you
just have parenthesis with absolutely nothing in between.
Because the star can include, including no repetition at all.
Remember the plus sign includes at least one of the item.
So, for example, here we're looking for at least one number,
followed by any number of characters, followed by at least one number again.
So here we see one number followed by some large number of characters,
followed by a number, another number again.
We see this again here, and so this allows us to look for any
possible combination of numbers that are separated by something other than numbers.
We can use the curly brackets as interval qualifiers to let us
specify both the minimum and the maximum number of times to match an expression.
So here what we're looking for is Bush, either capital or
lower case, and at the end debate.
And in-between what we're looking for is at least one space,
followed by something that's not a space, followed by at least one space.
And we want to see that between one and five times.
In other words we want to see something like space word space,
up between one and five times.
So that will match all of these because, in this case,
there are five words that occur between Bush and debates.
In this case there are three words that occur between Bush and
debates, and so forth.
So as long as there's between one and five word-like objects between the two,
we'll discover it with this regular expression.
If you use m, n within the curly bracket, that means at least m times, but
not more than n matches.
If you only use one number m, it means exactly that many matches.
If you use m comma, that means at least m matches to that regular expression.
In most implementations of regular expressions, the parentheses not only
limit the scope of the alternatives provided by this or, but
they can also be used to remember the tax, ta, text matched by the subexpression.
We refer to the matched text with escaped numbers, like escape 1,
escape 2, and so forth.
So, for example, here what we're looking for
is a space followed by some, at least one but possibly more characters.
Followed by at least one space,
followed by the exact same match that we saw within the parentheses.
So, for example, here we see, night space night, which will match this part,
plus a space and then a replication of the exact same phrase.
So that'll happen again with, blah blah blah, or so so itchy, and so forth.
So that's looking for repetition of a particular phrase.
The star is greedy, so it always matches the longest possible string that
satisfies a regular expression.
So in this case, we're starting at the beginning of a string, and
we're looking for an s.
Followed by some possibly large number of characters, followed by another s.
And so it's going to match the entire phrase here, because it goes for
the longest possible string that satisfies the expression.
And that's true for all of these phrases because they all start with an s and
end with an s.
[SOUND] If we want to make it less greedy, then what we can do
is we can actually use this question mark to change the greediness of the star.
So now we start with the s, now we follow up with some smaller number of characters.
It's not going to try to find the maximum length of the string.
Followed by something with an s at the end of the string.
So, in summary, regular expressions are used in many different languages,
not just in R.
They're composes of literals, that's the actual text that you're looking for.
And metacharacters, which allows us to look at alternatives or
look at specific places in a line, or
look at places where there are spaces, and find classes of characters or words.
Text processing via regular expression is particular powerful when you're
dealing with sort of unfriendly sources.
So sources where the data maybe are in a free text format or in a structured data
format where you're trying to just get access to a particular phrase.
They're typically used with the functions grep, grepl, sub, and gsub, like we've
talked about when we're searching for strings or trying to do some kind of
text processing to either the variable names or the variables themselves.
I'd like to thank Mark Hansen for some of the lecture material in this lecture.
For Roger Payne for putting it together.
