Final Exam Model Paper 6

i. fares.TAXI is the name TAXI in the module’s own box; from fares import bus copied the name bus into the program, so it can be used without the dot:

15001800

(3 × 600 = 1,800.) Then line 6 stops the program.

ii. import fares brings in only the module’s name, fares: TAXI stays in the module’s own namespace. Only bus was copied into the program’s names, so line 6 gives NameError: name 'TAXI' is not defined. Write print(fares.TAXI) (or import it too: from fares import bus, TAXI).

• i.It doesn’t try every lane in every town. It works out a lot beforehand: before anyone asks, the company’s computers find the important long-distance roads and store shortcuts along them, so a long route only searches the small roads near the start and the end and joins them to the stored shortcuts. (It also searches toward the goal rather than in every direction.)
• ii.A search can only use the graph it has. If a road is closed, a bridge washed away, or a new road missing from the map data, the route is wrong, because a route is only as right as the data; the map must be kept up to date.

SELECT country, COUNT(*)FROM permitsWHERE trek = 'Manang'GROUP BY countryORDER BY COUNT(*) DESC;

WHERE keeps only the Manang rows; GROUP BY makes one group per country, and COUNT(*) counts the rows in each.

i.

import pandas as pd homes = pd.read_csv("homestays.csv")print(len(homes))lamjung = homes[homes["district"] == "Lamjung"]print(lamjung["price"].mean())print(homes["price"].min())

ii. Lamjung’s two homestays cost 2,200 and 1,700, so their average is 1,950. mean() always gives a decimal number:

61950.01500

• i.This is rubber duck debugging. To explain each line out loud he must say it exactly, and then he notices the step he had skipped or the thing he had assumed without checking. The listener doesn’t have to answer: the explaining does the work, so a duck (or a pencil and paper) is enough.
• ii.For each line, what he expected it to do (which values it should hold, what it should print) and what actually happens; where the two differ, he has found the line to look at.

File exists: there is already a folder called treks in his home folder, so mkdir can’t make another. Nothing is wrong: he can go straight in with cd treks.
can't open file: this message comes from Python itself, which couldn’t find the program: it looked for /home/pemba/permits.py, in the folder he is in, but permits.py is in treks. He should go where the file is, cd treks, then run python3 permits.py.

• i.No. After training, the model’s parameters are frozen: it doesn’t learn from conversations, and a new chat starts with the model exactly as it was. This chatbot keeps notes about its user, a record kept outside the model, and adds them to the context of each new chat, so the model can read “sells pashmina shawls in Lakeside” there.
• ii.Because the notes are a record outside the model, she can usually see and delete them in the chatbot’s settings (or turn the notes off), and then new chats won’t have them.

Her test can’t show that. A swap test must change one detail only. She changed three at once (boy or girl, Kathmandu or Humla, maths or reading), so she can’t tell which one made the answers differ; liking reading could explain different careers on its own. And she asked each version only once, but a chatbot’s answers vary every time, so one difference may be chance. A fair test: the same sentence with only the name and “boy” or “girl” changed, each asked several times in new chats, and look for differences that appear every time.

An invoice is a bill she sends the client for each payment: what was done, the price, and how to pay. Kept with the bank’s record of the money arriving, it shows that the money is foreign income for her work, which she can show to get a loan, for a visa’s tax clearance, to pay the right tax, or if there is a dispute with the client.

The claim is wrong. First, subscriptions are not people: one person with two SIM cards and a home connection counts three times, while many others have none. In 2021 Nepal’s telecom regulator counted internet subscriptions for about 91 in every 100 people, but the International Telecommunication Union estimated that about 56 in every 100 people used the internet in 2023. Second, even people who are online don’t gain equally (inequity in outcomes): a weak signal, data that runs out, poor English or little skill means far less benefit than a fast connection and good English. The divide isn’t closed; to know who is online, a survey must count people, and then ask what they can do with it.

i. Any two: sell changes a global (total), so a wrong total could come from any line that touches it, and the function can’t be tested on its own; it prints instead of returning the price, so no other program can use the price (a side effect mixed with the work); the numbers 1200, 4 and 200 are written inside the function, not as named constants.

ii.

"""Jeep fares from Beni to Jomsom.""" PER_SEAT = 1200      # rupeesGROUP_SEATS = 4      # this many seats or more get the discountDISCOUNT = 200       # rupees off a group's price  def price(seats):    """The price in rupees for this many seats, Rs 200 off for 4 or more."""    total = seats * PER_SEAT    if seats >= GROUP_SEATS:        total = total - DISCOUNT    return total  if __name__ == "__main__":    # tests at the edge of the discount    print(price(3), "should be 3600")    print(price(4), "should be 4600")

iii.

import jeep total = 0for seats in [2, 5]:    p = jeep.price(seats)    print("Rs", p)    total = total + pprint("Total", total)

Rs 2400Rs 5800Total 8200

Now price is a pure function: the total is kept in the program’s own lines, where you can see every change, and the module can be tested and used by another program, such as a booking website.

i.

• ii.5 trails. Manang was first reached from Braka, Braka from Humde, Humde from Lower Pisang, Lower Pisang from Bhratang, Bhratang from Chame: Chame → Bhratang → Lower Pisang → Humde → Braka → Manang.
• iii.The fewest from Chame to Ghyaru is 3 (from the table), and from Ghyaru to Manang 3 (Ghyaru → Ngawal → Braka → Manang), so 6 trails: Chame → Bhratang → Upper Pisang → Ghyaru → Ngawal → Braka → Manang, one more than the fewest (5).

• i.The two guides from Solukhumbu: Pasang Sherpa and Mingma Lama.
• ii.One row per route, with its trekkers added up, the most first:

iii. The two Annapurna Circuit treks, each joined to its guide’s row, most trekkers first:

iv. Treks 2 and 7 still have guide_id 4. With his row gone they would point to a guide who doesn’t exist, and the JOIN in iii would quietly lose the treks he led. A foreign key rule on treks.guide_id makes the database refuse to delete a guide while any trek still points to him: keeping every foreign key pointing to a real row is referential integrity. (The treks must keep their history, so his row should stay.)

• i.(9, 3): the table has 9 rows (groups) and 3 columns (trek, people, days).
• ii.One row per trek, in alphabetical order, with the number of groups and the mean and median of their sizes:

                   count  mean  mediantrek                                  Annapurna Circuit      3  3.00     3.0Mardi Himal            4  7.25     3.5Upper Mustang          2  2.50     2.5

• iii.The median, 3.5. Mardi Himal’s groups had 4, 3, 20 and 2 people: one group of 20 (perhaps a school trip) pulls the mean up to 7.25, larger than three of the four groups. The median, the middle of the sizes in order, isn’t moved by one unusual value.
• iv.sort_values gives back a new, sorted table and leaves groups unchanged; line 7 throws the new one away. It should keep it: groups = groups.sort_values("days") (or print it straight away).

• i.Software as a service (SaaS): the agencies use a finished program in a browser, and look after only their own data; everything else is done for them.
• ii.Infrastructure as a service (IaaS): it rents computers (virtual machines) with storage and a network, and looks after the operating system, its programs and the data itself.
• iii.Latency is the time a message takes to arrive. Messages travel through fiber-optic cables at a fixed speed, so the farther away the data center, the longer each message takes there and back. Only bringing the computer closer makes it shorter.
• iv.Any one: if the internet or the company’s service goes down, no permits can be issued anywhere (everything depends on one company); the data of every trekker is kept by a company abroad, under another country’s laws; costs can grow, and moving to another company later is hard.

"""The rafting company's prices.""" PER_PERSON = {"Trishuli": 2500, "Seti": 3500, "Bhote Koshi": 4500}   # rupeesGROUP = 6            # this many people or more get the discountGROUP_OFF = 0.10     # 10% off  def trip_price(river, people):    """The price in rupees of a trip for this many people on this river.    Raises ValueError for a river the company doesn't run."""    if river not in PER_PERSON:        raise ValueError("No trips on " + river)    price = PER_PERSON[river] * people    if people >= GROUP:        price = price * (1 - GROUP_OFF)    return round(price)  if __name__ == "__main__":    # tests: either side of the group discount, and a river it doesn't run    print(trip_price("Seti", 5), "should be 17500")    print(trip_price("Seti", 6), "should be 18900")    try:        trip_price("Arun", 2)        print("No error: wrong!")    except ValueError:        print("Arun refused, as it should be")

Run by itself, rafting.py prints its tests:

17500 should be 1750018900 should be 18900Arun refused, as it should be

import rafting for group in range(1, 4):    river = input(f"Group {group}: which river? ")    people = int(input("How many people? "))    try:        print("Price: Rs", rafting.trip_price(river, people))    except ValueError:        print("No trips on that river")

Group 1: which river? TrishuliHow many people? 4Price: Rs 10000Group 2: which river? Bhote KoshiHow many people? 6Price: Rs 24300Group 3: which river? ArunHow many people? 3No trips on that river

4 × Rs 2,500 = Rs 10,000. 6 × Rs 4,500 = Rs 27,000, less 10% = Rs 24,300. Any clear names and messages are fine; a dictionary or if/elif for the prices are both right.

The same program, with these three lines in place of the last one:

import sqlite3 db = sqlite3.connect(":memory:")db.execute("CREATE TABLE homestays (homestay_id INTEGER PRIMARY KEY, owner TEXT, village TEXT, district TEXT, rooms INTEGER)")db.execute("""CREATE TABLE bookings (booking_id INTEGER PRIMARY KEY,              homestay_id INTEGER REFERENCES homestays(homestay_id),              country TEXT, guests INTEGER, nights INTEGER)""")db.executemany("INSERT INTO homestays VALUES (?, ?, ?, ?, ?)", [    (1, "Bishnu Gurung", "Sirubari", "Syangja", 3),    (2, "Maya Gurung", "Ghale Gaun", "Lamjung", 4),    (3, "Tika Magar", "Bandipur", "Tanahun", 2),    (4, "Lalita Tamang", "Briddim", "Rasuwa", 3),    (5, "Phurba Sherpa", "Thame", "Solukhumbu", 2),])db.executemany("INSERT INTO bookings VALUES (?, ?, ?, ?, ?)", [    (1, 2, "Japan", 2, 3),    (2, 1, "Nepal", 4, 1),    (3, 4, "Germany", 2, 5),    (4, 2, "India", 3, 2),    (5, 3, "Japan", 1, 4),    (6, 2, "Germany", 2, 6),    (7, 5, "Japan", 2, 2),    (8, 4, "Nepal", 5, 1),    (9, 1, "India", 2, 2),    (10, 4, "Japan", 3, 3),])  def ask(sql):    """Print each row the query gives, then a blank line."""    for row in db.execute(sql):        print(row)    print()  ask("SELECT booking_id, country, nights FROM bookings WHERE nights >= 4 ORDER BY nights DESC")ask("SELECT homestay_id, SUM(nights) FROM bookings GROUP BY homestay_id HAVING SUM(nights) > 5")ask("""SELECT homestays.village, bookings.guests FROM bookings       JOIN homestays ON bookings.homestay_id = homestays.homestay_id       WHERE bookings.country = 'Japan' ORDER BY homestays.village""")

(6, 'Germany', 6)(3, 'Germany', 5)(5, 'Japan', 4) (2, 11)(4, 9) ('Bandipur', 1)('Briddim', 3)('Ghale Gaun', 2)('Thame', 2)

(The bookings are made up for this paper.)

flights = {    "Kathmandu": ["Pokhara", "Lukla", "Nepalgunj", "Biratnagar", "Tumlingtar", "Bharatpur"],    "Pokhara": ["Kathmandu", "Jomsom", "Nepalgunj"],    "Jomsom": ["Pokhara"],    "Lukla": ["Kathmandu"],    "Nepalgunj": ["Kathmandu", "Pokhara", "Simikot", "Jumla", "Surkhet"],    "Simikot": ["Nepalgunj"],    "Jumla": ["Nepalgunj", "Surkhet"],    "Surkhet": ["Nepalgunj", "Jumla"],    "Biratnagar": ["Kathmandu", "Tumlingtar"],    "Tumlingtar": ["Kathmandu", "Biratnagar"],    "Bharatpur": ["Kathmandu"],    "Manang": [],}  def fewest_flights(graph, start, goal):    """The airports on a route with the fewest flights, or None."""    came_from = {start: None}    to_visit = [start]    while len(to_visit) > 0:        airport = to_visit.pop(0)        if airport == goal:            route = []            while airport is not None:                route.append(airport)                airport = came_from[airport]            route.reverse()            return route        for nxt in graph[airport]:            if nxt not in came_from:                came_from[nxt] = airport                to_visit.append(nxt)    return None  for start, goal in [("Jomsom", "Simikot"), ("Lukla", "Jumla"), ("Tumlingtar", "Manang")]:    route = fewest_flights(flights, start, goal)    if route is None:        print(start, "to", goal + ": no flights")    else:        print(start, "to", goal + ":", " → ".join(route), "(" + str(len(route) - 1), "flights)")

Jomsom to Simikot: Jomsom → Pokhara → Nepalgunj → Simikot (3 flights)Lukla to Jumla: Lukla → Kathmandu → Nepalgunj → Jumla (3 flights)Tumlingtar to Manang: no flights

Jomsom to Simikot through Kathmandu would be 4 flights. Any route with the fewest flights is right.

• i.A price of 0 would look like a real answer: main.py could print “Price: Rs 0” and a group might think the trip is free. An error can’t be mistaken for a price, and the caller must decide what to do. main.py catches it with try / except ValueError and prints “No trips on that river”, so the program goes on to the next group instead of stopping.
• ii.“4 nights or more” includes 4: > 4 would leave out booking 5, of exactly 4 nights. That is the edge, where mistakes hide. 'Japan' is text, which SQL writes in quotes; nights is an INTEGER field, and numbers are written without quotes.
• iii.Its keys are the airports seen so far, and each value is the airport it was first reached from. Following these notes from the goal leads back one airport at a time to the start (whose note is None), so the route is collected goal first; reverse() puts the start first.