I have a relation Presidents(firstName,lastName,beginTerm,endTerm)
that gives information about US Presidents. Attribute firstName is a string
with the first name, and in some cases, one or more
middle initials.
Attribute lastName is a string with the last name of the president. For example,
the previous president has firstName = 'George W.' and his father has firstName = 'George H.W.'; both have lastName = 'Bush'. The last 2 attributes, beginTerm and endTerm,
are the years the president entered and left office, respectively.
One subtlety is that Grover Cleveland served 2 noncontiguous
terms. He appears in 2 tuples, one with the beginning and ending years of his first term and the other for the second term.
The question I have is below:
There are 2 pairs of presidents that were father and son. But there are
a number of other pairs of presidents that shared a last name. Find all the last names belonging to 2 or more Presidents. Do not repeat a last name, and remember that the same person serving 2 different terms (e.g., Grover Cleveland) does not constitute a case of 2 presidents with the same last name.
I first thought the answer might be:
SELECT lastName
FROM Presidents
WHERE COUNT(lastName) > 2
EXCEPT lastName = 'Cleveland';
I'm not too sure if the COUNT() function can be used in the WHERE clause though.
Is this possible?
Thanks!
Use HAVING instead of WHERE when checking against Group functions.
SELECT lastName
FROM Presidents
WHERE lastName != 'Cleveland'
GROUP BY lastName
HAVING COUNT(lastName) > 2;
However, when solving SQL-puzzles likes this, you should never take into account the actual data. It should work for all consistent data-sets! I believe this is an actual solution to your problem:
SELECT DISTINCT p1.lastName
FROM Presidents p1, Presidents p2
WHERE p1.lastName == p2.LastName
AND p1.firstName != p2.firstName;
You constrain on aggregates using HAVING, and you are also missing a group by.
SELECT lastName
FROM Presidents
where lastName <> 'Cleveland';
group by lastname
having COUNT(lastName) > 2
Assuming there is an id field as well,
select id, lastname, count(*) differentguycount
from presidents left join
(select id, firstname, lastname count(*) sameguycount
from presidents
group by id, firstname, lastname
having sameguycount > 1 ) temp on temp.id = presidents.id
where temp.firstname is null
group by id, lastname
having differentguycount > 1
As noted, the OP did not specify his database engine which could cause syntax errors. For example some databases might not allow you to use aliases in the having clause.
Related
I have a table of invoice data with over 100k unique invoices and several thousand unique company names associated with them.
I'm trying to group these company names into more general groups to understand how many invoices they're responsible for, how often they receive them, etc.
Currently, I'm using the following code to identify unique company names:
SELECT DISTINCT(company_name)
FROM invoice_data
ORDER BY company_name
The problem is that this only gives me exact matches, when its obvious that there are many string values in company_name that are similar. For example: McDonalds Paddington, McDonlads Oxford Square, McDonalds Peckham, etc.
How can I make by GROUP BY statement more general?
Sometimes the issue isn't as simple as the example listed above, occasionally there is simply an extra space or PTY/LTD which throws off a GROUP BY match.
EDIT
To give an example of what I'm looking for, I'd be looking to turn the following:
company_name
----------------------
Jim's Pizza Paddington|
Jim's Pizza Oxford |
McDonald's Peckham |
McDonald's Victoria |
-----------------------
And be able to group by their company name rather than exclusively with an exact string match.
Have you tried using the Soundex function?
SELECT
SOUNDEX(name) AS code,
MAX( name) AS sample_name,
count(name) as records
FROM ((
SELECT
"Jim's Pizza Paddington" AS name)
UNION ALL (
SELECT
"Jim's Pizza Oxford" AS name)
UNION ALL (
SELECT
"McDonald's Peckham" AS name)
UNION ALL (
SELECT
"McDonald's Victoria" AS name))
GROUP BY
1
ORDER BY
You can then use the soundex to create groupings, with a split or other type of function to pull the part of the string which matches the name group or use a windows function to pull back one occurrence to get the name string. Not perfect but means you do not need to pull into other tools with advanced language recognition.
We've had a few cases of people entering in first names where last names should be and vice versa. So I'm trying to come up with a SQL search to match the swapped columns. For example, someone may have entered the record as first_name = Smith, last_name = John by accident. Later, another person may see that John Smith is not in the database and enter a new user as first_name = John, last_name = Smith, when in fact it is the same person.
I used this query to help narrow my search:
SELECT person_id, first_name, last_name
FROM people
WHERE first_name IN (
SELECT last_name FROM people
) AND last_name IN (
SELECT first_name FROM people
);
But if we have people named John Allen, Allen Smith, and Smith John, they would all be returned even though none of those are actually duplicates. In this case, it's actually good enough that I can see the duplicates in my particular data set, but I'm wondering if there's a more precise way to do this.
I would do a self join like this:
SELECT p1.person_id, p1.first_name, p1.last_name
FROM people p1
join people p2 on p1.first_name = p2.last_name and p1.last_name = p2.first_name
To also find typos on names I recommend this:
SELECT p1.person_id, p1.first_name, p1.last_name
FROM people p1
join people p2 on soundex(p1.first_name) = soundex(p2.last_name) and
soundex(p1.last_name) = soundex(p2.first_name)
soundex is a neat function that "hashes" words in a way that two words that sound the same get the same hash. This means Anne and Ann will have the same soundex. So if you had an Anne Smith and a Smith Ann the query above would find them as a match.
Interesting. This is a problem that I cover in Data Analysis Using SQL and Excel (note: I only very rarely mention books in my answers or comments).
The idea is to summarize the data to get a likelihood of a mismatch. So, look at the number of times a name appears as a first name and as a last name and then combine these. So:
with names as (
select first_name as name, 1.0 as isf, 0.0 as isl
from people
union all
select last_name, 0, 1
from people
),
nl as (
select name, sum(isf) as numf, sum(isl) as numl,
avg(isf) as p_f, avg(isl) as p_l
from names
group by name
)
select p.*
from people p join
nl nlf
on p.first_name = nlf.name join
nl nll
on p.last_name = nll.name
order by (coalesce(nlf.p_l, 0) + coalesce(nll.p_f, 0));
This orders the records by a measure of mismatch of the names -- the sum of the probabilities of the first name used by a last name and a last name used as a first name.
I want to find how many modules a lecturer taught in a specific year and want to select name of the lecturer and the number of modules for that lecturer.
Problem is that because I am selecting Name, and I have to group it by name to make it work. But what if there are two lecturers with same name? Then sql will make them one and that would be wrong output.
So what I really want to do is select name but group by id, which sql is not allowing me to do. Is there a way around it?
Below are the tables:
Lecturer(lecturerID, lecturerName)
Teaches(lecturerID, moduleID, year)
This is my query so far:
SELECT l.lecturerName, COUNT(moduleID) AS NumOfModules
FROM Lecturer l , Teaches t
WHERE l.lecturerID = t.lecturerID
AND year = 2011
GROUP BY l.lecturerName --I want lectureID here, but it doesn't run if I do that
SELECT a.lecturerName, b.NumOfModules
FROM Lecturer a,(
SELECT l.lecturerID, COUNT(moduleID) AS NumOfModules
FROM Lecturer l , Teaches t
WHERE l.lecturerID = t.lecturerID
AND year = 2011
GROUP BY l.lecturerID) b
WHERE a.lecturerID = b.lecturerID
You should probably just group by lecturerID and include it in the select column list. Otherwise, you're going to end up with two rows containing the same name with no way to distinguish between them.
You raise the problem of "wrong output" when grouping just by name but "undecipherable output" is just as big a problem. In other words, your desired output (grouping by ID but giving name):
lecturerName Module
------------ ------
Bob Smith 1
Bob Smith 2
is no better than your erroneous output (grouping by, and giving, name):
lecturerName Module
------------ ------
Bob Smith 3
since, while you now know that one of the lecturers taught two modules and the other taught one, you have no idea which is which.
The better output (grouping by ID and displaying both ID and name) would be:
lecturerId lecturerName Module
---------- ------------ ------
314159 Bob Smith 1
271828 Bob Smith 2
And, yes, I'm aware this doesn't answer your specific request but sometimes the right answer to "How do I do XYZZY?" is "Don't do XYZZY, it's a bad idea for these reasons ...".
Things like writing operating systems in COBOL, accounting packages in assembler, or anything in Pascal come to mind instantly :-)
You could subquery your count statement.
SELECT lecturername,
(SELECT Count(*)
FROM teaches t
WHERE t.lecturerid = l.lecturerid
AND t.year = 2011) AS NumOfModules
FROM lecturer l
Note there are other ways of doing this. If you also wanted to elimiate the rows with no modules you can then try.
SELECT *
FROM (SELECT lecturername,
(SELECT Count(*)
FROM teaches t
WHERE t.lecturerid = l.lecturerid
AND t.year = 2011) AS NumOfModules
FROM lecturer l) AS temp
WHERE temp.numofmodules > 0
I have a table of listings that has NAP fields and I wanted to find duplication within it - specifically where everything is the same except the house number (within 2 or 3 digits).
My table looks something like this:
Name Housenumber Streetname Streettype City State Zip
1 36 Smith St Norwalk CT 6851
2 38 Smith St Norwalk CT 6851
3 1 Kennedy Ave Campbell CA 95008
4 4 Kennedy Ave Campbell CA 95008
I was wondering how to set up a qry to find records like these.
I've tried a few things but can't figure out how to do it - any help would be appreciated.
Thanks
Are you looking to find something that shows the amount of these rows you have like this?
SELECT
StreenName,
City,
State,
Zip,
COUNT(*)
FROM YourTable
group by StreenName, City, State, Zip
HAVING COUNT(*) >1
Or maybe trying to find all of the rows that have the same street, city, state, and zip?
SELECT
A.HouseNumber,
A.StreetName,
A.City,
A.State,
A.Zip
FROM YourTable as A
INNER JOIN YourTable as B
ON A.StreetName = B.StreetName
AND A.City = B.City
AND A.State = B.State
AND A.Zip = B.Zip
AND A.HouseNumber <> B.HouseNumber
Here is one way to do it. You'll need a unique ID for the table to run this, as you wouldn't want to select the exact same person if theyre the only one there. This'll just spit out all the results where there is at least one duplicate.
Edit: Woops, just realized in comments it says varchar for the street number...hmm. So you could just run a cast on it. The OP never said anything about house numbers in varchar or being letters and numbers in the original post. As for letters in the street number field, I've been a third party shipping provider for 2 yrs in the past and I have never seen one; with the exception of an apt., which would be a diff field. Its just as likely that someone put varchar there for some other reason(leading 0's), or for no reason. Of oourse there could be, but no way of knowing whats in the field without response from OP. To run cast to int its the same except this for each instance: Cast(mt.HouseNumber as int)
select *
from MyTable mt
where exists (select 1
from MyTable mt2
where mt.name = mt2.name
and mt.street = mt2.street
and mt.state = mt2.state
and mt.city = mt2.city
and mt2.HouseNumber between (mt.HouseNumber -3) and (mt.HouseNumber +3)
and mt.UID != mt2.UID
)
order by mt.state, mt.city, mt.street
;
Not sure how to run the -3 +3 if there are letters involed...unless you know excatly where they are and you can just simply cut them out then cast.
I have a table called babynames that looks abit like this:
firstname |sex |year |count
Bob |M| 2010| 150
Bob |M| 2009| 100
Bob |M| 2008| 122
Bob |F| 2007| 2
Bob |F| 2001| 1
What I want to do is get a list of all the baby names that are both female and male, so my query needs to pull all the firstname records that have at least two records in the table and are at least one M and one F.
It's getting late and my mind isn't working well tonight. Can anyone suggest a string that might help me achieve this task?
There are several ways to handle this. One would be to to use a COUNT(DISTINCT sex) = 2 in the HAVING clause. Be sure to GROUP BY firstname.
SELECT
firstname
FROM babynames
GROUP BY firstname
HAVING COUNT(DISTINCT sex) = 2
Here's a demo: http://sqlfiddle.com/#!2/5d221/1
Another would an INNER JOIN against 2 aliases of the same table, where one looks for M while the other looks for F. If a name isn't matched by both conditions, the join can't be made and it will get excluded from the output.
SELECT
DISTINCT
m.firstname
FROM
babynames f
INNER JOIN babynames m ON f.firstname = m.firstname
WHERE
f.sex = 'F'
AND m.sex = 'M'
http://sqlfiddle.com/#!2/5d221/3