Description
Part 1: Text, Joins and Subqueries
For some reason, your instructor has been scraping the Caltrans website every 15 minutes or so, since
2015, to get road conditions on all of the highways within California. The data is written to MySQL.
Your version of the data is hourly, and only for 2017.
A Caltrans highway conditions report looks like the following and contains conditions for individual
stretches of highway (“area”) typically representing a coarse area of the state: Northern, Southern, Central,
Sierra Nevada etc.
SR 120
[IN THE CENTRAL CALIFORNIA AREA & SIERRA NEVADA]
IS CLOSED FROM CRANE FLAT TO 5 MI WEST OF THE JCT OF US 395 /TIOGA PASS/
(TUOLUMNE, MONO CO) – FOR THE WINTER – MOTORISTS ARE ADVISED TO USE AN
ALTERNATE ROUTE
[YOSEMITE NAT’L PARK]
FOR YOSEMITE NAT’L PARK ROAD INFORMATION CALL 209-372-0200
The schema for the caltrans table looks like the following:
CREATE TABLE ‘caltrans‘ (
reported TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
highway VARCHAR(6) NOT NULL,
area VARCHAR(255),
text TEXT NOT NULL,
hash VARCHAR(32) NOT NULL
);
1
reported is the time the data was scraped, highway is the highway the status pertains to prefixed by its
type (i.e. US101, SR1, I405), area refers to a particular part of the state or highway, and text is the update
itself. Since we cannot use text as a primary key, a hash column was added.
US Highway (US) California State Route (SR) Interstate (I)
Exercises
(a) Write a query that returns a list of all the highways in 2017 that were closed due to snow at any point of the
year, or were closed for the winter. Order them by highway and area and give us the top 20 results, both
columns in descending order. (Hint 1: You don’t need to do anything with dates to answer this question.
Hint 2: Before writing a query, look at the data.)
(b) For each highway found in part (a), compute the percentage of days out of the year that the highway was
closed. If a highway was closed for only a partial day, it counts as a full day. There are at least three ways
to solve this problem. Try to use a method that involves a join, and a method that does not. Report the
highway, area/stretch, the percentage of days it was closed in descending order by percentage, and only gives
us the 5 highest percentages and the highways and areas they belong to. You may hardcode the number of
days in the year (see the note below).
Note that not all of your responses with be perfect in theory because there were times when the instructor’s
script lost Internet connectivity, or the power went out. So, there may not be exactly 365 days in 2017, but
we can just assume it for simplicity. You can also try to find the number of days represented in this dataset
yourself if you wish, and use that number instead. Of course, if you want to write much more complicated
queries, you can use a subquery to compute the number of days, but in this particular case it is a rabbit
hole.
Part 2: Join Definitions
Exercises
(a) Your instructor almost included the above Venn Diagram in his lecture slides to show how different types
of joins are related, but he noticed that it was wrong in at least one way. Explain at least one thing that is
wrong about the diagram.
2
Part 3: More Joins and Subqueries
In Homework 1, we did several things with the Bird Scooter use case, but we did not have any data to practice
writing queries with. Suppose we now have trip data in the following two tables:
1. trip starts;
2. trip ends;
Exercises
(a) Write a query that computes the elapsed time of each trip. If something happened and a trip end was not
recorded, the elapsed time shall be 24 hours, per Bird’s policy. Print your results as trip id, user id, and
trip length. Only show the first 5, without any special ordering.
(b) Write a query that computes the charge to the user for each trip. The charge is calculated as follows: $1 flat
rate per trip plus 15 cents per minute. All fractional minutes are rounded up to the next minute. Assume
we did not store the results of the query from part (a). Print the first 5 results (no ordering) as trip id,
user id and trip charge.
(c) Putting it all together: Suppose we bill the user at the end of the month rather than at the end of each trip.
Write a query that computes the monthly charge for trips in March 2018 for each user assuming we did not
store the results from parts (a) or (b). Print your results: user id and monthly total for the first five users
(no ordering). In particular, how much does user id = 2 owe?
(d) In the solution set for Homework 1, it was mentioned that another way we can record starts and ends of
trips was to use one table, 2 rows per trip: one row representing the start and a second row representing the
end of the trip. We would then have an ENUM or BIT that specifies whether the row refers to a start or an
end. If we wanted to use this one single table as the basis to charge users, what type of join would we need
to compute?
3
