Markscheme: Practice Paper 1, SL, set 2

Mark your own answers point by point. The number in brackets after each point is what it is worth.

Section A

Answer all questions.

Question 1

A school's lighting controller stores each room's brightness level as one unsigned byte.

(a) Convert the denary value 183 into 8-bit binary. [1]

(b) Convert the denary value 183 into hexadecimal. [1]

Show your working.

[3]

p1 [1]
(a) Gives 10110111 (128 + 32 + 16 + 4 + 2 + 1 = 183).
p2 [1]
(b) Gives B7 (1011 is B and 0111 is 7, or 183 = 11 x 16 + 7).
p3 [1]
(c) Gives 62 (3 x 16 + 14 = 62).

Also accept

For p2, "0xB7" or "b7" earns the mark.
For p1, "1011 0111" written with a space earns the mark.
Follow through for p2 from a wrong binary answer in (a) if the nibbles are converted correctly.

No marks for

For (b), "117". That reads the nibbles 1011 and 0111 as the denary numbers 11 and 7 side by side.
For (c), "314". That treats E as 14 and writes 3 and 14 next to each other instead of multiplying the 3 by 16.

A full-marks answer

(a) 183 = 128 + 32 + 16 + 4 + 2 + 1, so the binary is 10110111.

(b) Split into nibbles: 1011 = 11 = B and 0111 = 7, so 183 is B7 in hexadecimal.

Question 2

A wildlife camera in a nature reserve saves uncompressed bitmap photos that are 1024 pixels wide and 768 pixels high, with a colour depth of 24 bits.

(a) Calculate the size of one photo in bytes. Ignore any file header. [2]

(b) To save space, the colour depth is reduced to 16 bits. State how many different colours each pixel can then show. [1]

[3]

p1 [1]
(a) Works out the number of pixels times the bits (or bytes) per pixel, for example 1024 x 768 x 24 = 18 874 368 bits, or 24 bits = 3 bytes per pixel.
p2 [1]
(a) Gives 2 359 296 bytes (1024 x 768 x 3).
p3 [1]
(b) States 65 536 colours (2 to the power 16).

Also accept

A correct final answer of 2 359 296 bytes with no working earns both p1 and p2.
For p3, "2^16" earns the mark.

No marks for

For (a), 18 874 368 given as the answer in bytes. That is the number of bits; it has not been divided by 8.
For (b), 16 colours, or 32 colours (16 x 2).

A full-marks answer

(a) Pixels: 1024 x 768 = 786 432. Each pixel uses 24 bits = 3 bytes. Size = 786 432 x 3 = 2 359 296 bytes.

(b) 2^16 = 65 536 colours.

Question 3

A bike-share docking station stands outside all year. Its small computer stores the station's software and a log of hires, which is a few gigabytes in total.

Explain why a solid-state drive (SSD) is a better choice than a hard disk drive (HDD) for this computer. [3]

p1 [1]
An SSD has no moving parts, so it is more durable and less likely to fail from knocks, vibration or temperature changes outdoors.
p2 [1]
An SSD has faster access times, so the station's software starts and the hire log is read or written quickly.
p3 [1]
Only a few gigabytes are needed, so the higher cost per gigabyte of an SSD (and the larger capacity of an HDD) does not matter here.

Also accept

For p1, "an HDD has a spinning disk and a moving read/write head that can be damaged" earns the mark.
For p3, "a small SSD is cheap enough" earns the mark if linked to the small amount of data.
An SSD being smaller and lighter, so it fits a small enclosure, earns p2 in place of speed if linked to the station.
Lower power use or less heat from an SSD earns p2 in place of speed, if explained for this station.

No marks for

Saying an SSD has more storage capacity than an HDD. For the same price, HDDs are usually larger.
Saying an SSD does not lose data when the power is off, as if an HDD does. Both are non-volatile.

A full-marks answer

An SSD has no moving parts, while an HDD has a spinning platter and a moving head, so an SSD copes much better with knocks, vibration and the cold and heat outside. An SSD also has faster access times, so the software starts quickly and each hire is logged quickly. The usual drawbacks of an SSD, a higher cost per gigabyte and smaller capacity, do not matter because the station only needs a few gigabytes.

Question 4

A student's laptop has three processes ready to run at time 0, in the order shown. Each needs the CPU for the number of time units in the table.

Process	CPU time needed
P1 (video export)	5
P2 (chat message)	2
P3 (spell check)	3

(a) Deduce the time at which each process finishes under first come first served (FCFS) scheduling. [1]

(b) Deduce the time at which each process finishes under round robin scheduling with a time slice of 2 units. A process that has not finished goes to the back of the queue. Show the order in which the processes use the CPU. [2]

[5]

p1 [1]
(a) P1 finishes at 5, P2 at 7 and P3 at 10.
p2 [1]
(b) Gives the order of time slices P1 (0 to 2), P2 (2 to 4), P3 (4 to 6), P1 (6 to 8), P3 (8 to 9), P1 (9 to 10), or a correct Gantt chart showing it.
p3 [1]
(b) P2 finishes at 4, P3 at 9 and P1 at 10.
p4 [1]
(c) Under FCFS the chat process has to wait for the whole video export to finish first (it finishes at 7, not 4), so the short task is held up by a long one.
p5 [1]
(c) Round robin gives every process a turn within a short, fixed time, so interactive tasks such as chat respond quickly and no process can hold the CPU for long.

Also accept

For p3, finish times read correctly from the student's own Gantt chart earn the mark if the chart only has a single slip in one slice.
For p5, "fairer, every process gets a share of CPU time in turn" earns the mark when linked to the chat feeling responsive.

No marks for

For (b), P2 finishing at 2. P1 runs first, for one full time slice.
For (c), "round robin makes every process finish sooner". P1 finishes at 10 under both, so the gain is a quicker response for short tasks, not a shorter total time.
For (c), "round robin runs the processes at the same time". On one core it switches between them.

A full-marks answer

(a) FCFS: P1 runs 0 to 5, P2 runs 5 to 7, P3 runs 7 to 10. P1 finishes at 5, P2 at 7, P3 at 10.

(b) Round robin, slice 2: P1 0 to 2 (3 left), P2 2 to 4 (done), P3 4 to 6 (1 left), P1 6 to 8 (1 left), P3 8 to 9 (done), P1 9 to 10 (done). P2 finishes at 4, P3 at 9 and P1 at 10.

(c) Under FCFS the chat message must wait behind the long video export and only finishes at 7. Round robin gives each process a short turn in rotation, so the chat process gets the CPU within 2 units and finishes at 4. No single long task can hold the CPU, so interactive programs stay responsive.

Question 5

A school has 400 devices that use private IP addresses, but the school has only one public IP address.

(a) Distinguish between an IP address and a MAC address. [2]

(b) State the technique the school router uses to let all 400 devices reach the internet through the one public IP address. [1]

[3]

p1 [1]
(a) A MAC address is assigned to the network interface card by its manufacturer (burned in), while an IP address is given to the device by the network it joins (by DHCP or by hand), so it changes when the device moves network.
p2 [1]
(a) A MAC address is used to deliver frames within the local network, while an IP address is used to route packets between networks.
p3 [1]
(b) States network address translation (NAT).

Also accept

For p1, "the MAC address is a physical address that comes with the card; the IP address is a logical address the network hands out" earns the mark.
For p1, noting that many phones now use a randomised MAC address on Wi-Fi does not lose the mark.
For p2, "MAC addresses are used on the LAN; IP addresses get data across the internet" earns the mark.
For p3, "port address translation (PAT)" or "NAT overload" earns the mark.

No marks for

For (a), "an IP address is for the internet and a MAC address is for Wi-Fi". Wired and wireless devices both have MAC and IP addresses.
For (a), "a MAC address is only on Apple computers". MAC stands for media access control.
For (b), "DHCP". DHCP hands out private addresses inside the school; it does not let them share the public one.

A full-marks answer

(a) A MAC address is assigned to the network interface card by its manufacturer, while an IP address is given to the device by the network it joins, for example by DHCP, so it changes when the device moves to another network. The MAC address is used to deliver frames between devices on the same local network, while the IP address is used to route packets across different networks.

(b) Network address translation (NAT).

Question 6

A school has four buildings, and each building has its own switch. The school is choosing how to cable the four building switches together. Option 1 is a star: a new core switch in the server room, with one cable to each building switch. Option 2 is a full mesh: a cable directly between every pair of building switches.

(a) Calculate the number of cables Option 2 needs. [1]

(b) Compare Option 1 and Option 2 in terms of cost and resilience. [2]

[3]

p1 [1]
(a) Gives 6 as the number of cables for the full mesh.
p2 [1]
(b) The star is cheaper, because it needs fewer cables (4 against 6).
p3 [1]
(b) The mesh is more resilient, because if one cable fails traffic can go another way, while in the star a failed core switch cuts every building off from the others (a single point of failure).

Also accept

For p1, the working 4 x 3 / 2 = 6 earns the mark.
For p2, "the star uses less cabling and fewer switch ports, so it costs less to install" earns the mark.
For p3, "in the star, a broken cable cuts off one building, but in the mesh there is always another route" earns the mark.

No marks for

For (a), 12. That counts each cable twice, once from each end.
For (b), "the mesh is better" or "the star is better" with no reference to cost or resilience. A comparison needs the reason.
For (b), "the star is more resilient because it has a central switch". The central switch is exactly the single point of failure.

A full-marks answer

(a) Each of the 4 switches links to the other 3, and each cable joins two switches, so 4 x 3 / 2 = 6 cables.

(b) The star is cheaper: it needs only 4 cables to the core switch against 6 for the mesh. The mesh is more resilient: if a cable fails, traffic can take another route between buildings, but in the star, if the core switch fails every building is cut off from the others, because it is a single point of failure.

Question 7

A student at home uploads a large photo of a science poster to the school's file server. The photo is sent across the internet using packet switching.

(a) State one item, other than part of the photo itself, that each packet carries. [1]

(b) Explain why the packets can arrive at the file server in a different order from the one they were sent in. [2]

[3]

p1 [1]
(a) Names one header or trailer item, such as the destination IP address, the source IP address, a sequence number, a time to live (hop limit), or a checksum.
p2 [1]
(b) Each router decides where to send each packet separately, based on current conditions such as congestion or failed links, so packets from one photo can take different routes.
p3 [1]
(b) Different routes have different delays, so a packet sent later can arrive before one sent earlier (the server then uses sequence numbers to put them back in order).

Also accept

For p1, "the address it is going to", "the packet's number in the sequence", "the protocol" or "an error check" earns the mark.
For p3, "some routes are busier or longer than others, so packets are overtaken" earns the mark.
For p3, "a lost packet is sent again, so it arrives after packets that were sent later" earns the mark.

No marks for

For (a), "the file name". That is part of the data being sent, not packet information.
For (b), "packets are sent in a random order". The sender sends them in order; the order changes on the way.
For (b), "the switch mixes them up". Packets cross the internet through routers, which choose routes independently.

A full-marks answer

(a) The destination IP address.

(b) Each router along the way decides independently where to forward each packet, using current conditions such as how busy each link is, so packets from the same photo can travel by different routes. Those routes have different delays, so a packet sent later can overtake one sent earlier; the file server uses the sequence numbers to rebuild the photo in the right order.

Question 8

A school has a firewall between its router and the internet. One morning a student plugs in a USB stick at a library computer, and malware on the stick spreads to other computers on the school network.

(a) Outline what the firewall does. [2]

(b) Explain why the firewall did not stop the malware from spreading. [2]

[4]

p1 [1]
(a) The firewall inspects the traffic passing between the school network and the internet (incoming, outgoing or both).
p2 [1]
(a) It compares each packet with a set of rules (for example about IP addresses, port numbers or protocols) and allows or blocks it.
p3 [1]
(b) The malware entered on a device already inside the network, so it never crossed the firewall, which only sees traffic at the boundary with the internet.
p4 [1]
(b) The spread happened between computers inside the school network, and that internal traffic does not pass through the firewall.

Also accept

For p2, "filters traffic using rules set by the network administrator" earns the mark.
For p3, "the USB stick bypassed the firewall" earns the mark if it says why (the stick is plugged in inside the network).
For p4, "traffic between two school computers goes through the school switches, not through the firewall" earns the mark.
For p4, "a packet filter checks headers (addresses, ports, protocols), not file contents, so it would not detect malware inside traffic its rules allow" earns the mark.

No marks for

For (a), "it scans files for viruses". That is antivirus software, not a firewall.
For (b), "the firewall was switched off" or "the rules were wrong". Nothing in the scenario says so, and correct rules would still not see this traffic.
For (a), "it stops all hackers". A firewall only applies its rules to the traffic it sees.

A full-marks answer

(a) The firewall inspects the traffic passing between the school network and the internet, and compares each packet with its rules, about things such as IP addresses, ports and protocols, to decide whether to allow or block it.

(b) The malware came in on a USB stick plugged into a computer already inside the network, so it never crossed the firewall at the internet boundary. It then spread from computer to computer inside the school network, and that internal traffic does not pass through the firewall at all.

Question 9

A school drama club stores all its data in one table, Casting.

student_id	student_name	role	play_title	play_date
11	Priya	Narrator	The Lighthouse	2026-12-04
12	Tom	Keeper	The Lighthouse	2026-12-04
13	Ines	Fox	Winter Market	2027-03-12
11	Priya	Baker	Winter Market	2027-03-12
14	Omar	Captain	Paper Boats	2027-05-20

(a) The Winter Market performance moves to 2027-03-19. Describe the update anomaly that can happen with this table. [2]

(b) Omar leaves the club and his record is deleted. Describe the deletion anomaly: what else the database loses, apart from Omar's own details. [1]

[3]

p1 [1]
(a) Says the date of Winter Market is stored in more than one record (two here), so the change has to be made in every one of them.
p2 [1]
(a) Says that if only some of those records are changed, the table holds two different dates for the same play, so the data is inconsistent.
p3 [1]
(b) Says the only record of the play Paper Boats and its date (2027-05-20) is deleted too, so the database no longer knows the play exists.

Also accept

For p1, "play_date is repeated for every cast member of the play" earns the mark.
For p2, "one record says 2027-03-12 and the other says 2027-03-19, so nobody knows which is right" earns the mark.

No marks for

For (a), "the update fails" or "the database gives an error". The update runs; the problem is that the result can disagree with itself.
For (a), describing an insertion anomaly (a new play cannot be added without a student). It is true but it is not what the question asks.
For (b), "Omar's data is lost". Losing Omar's record is the intended effect, not the anomaly.

A full-marks answer

(a) The date of Winter Market is stored twice, once in Ines's record and once in Priya's, so the new date has to be written into both records. (p1) If only one is changed, the table says the same play is on two different dates, so the data is inconsistent. (p2)

(b) Omar's record was the only one for Paper Boats, so deleting it also deletes the play's title and its date, 2027-05-20. The database no longer knows that play exists. (p3)

Question 10

A music school has drawn an entity relationship design for its lessons. These tables describe it.

Entity	Attributes
Student	student_id (primary key), name
Course	course_code (primary key), instrument, grade_level
Teacher	teacher_id (primary key), name

Relationship	Cardinality
Teacher teaches Course	one-to-many (each course has exactly one teacher)
Student enrols on Course	many-to-many

(a) Explain why the relationship between Student and Course cannot be put into tables by adding one foreign key to Student or to Course. [1]

(b) Construct a set of tables that puts this whole design into a relational database. For each table, give its attributes, its primary key and any foreign keys. [3]

[4]

p1 [1]
(a) Explains that one foreign key field holds one value, but a student takes many courses and a course has many students, so either table would need a list of values (a repeating group) in that field.
p2 [1]
(b) Adds a link table, for example Enrolment, holding student_id and course_code with the pair as its composite primary key.
p3 [1]
(b) States that in the link table student_id is a foreign key to Student and course_code is a foreign key to Course.
p4 [1]
(b) Puts teacher_id into Course as a foreign key to Teacher (on the many side), and adds no foreign key to Teacher.

Also accept

For p1, "a Student record would have to repeat once for every course, duplicating the student's data" earns the mark.
For p2, a link table with its own key (such as enrolment_id) earns the mark if the pair (student_id, course_code) is also said to be unique.
For p2, any sensible name for the link table, such as StudentCourse, earns the mark.

No marks for

For (a), "many-to-many is not allowed" with no reason.
For (b), putting course_code into Teacher. A teacher teaches many courses, so that is the same problem as in (a).
For (b), a link table between Teacher and Course. That relationship is one-to-many and needs only a foreign key.

A full-marks answer

(a) A foreign key field holds one value per record, but one student enrols on many courses and one course has many students, so whichever table got the foreign key would need many values in it, or the record would have to be repeated. (p1) (b) Student(student_id, name). Primary key student_id. Teacher(teacher_id, name). Primary key teacher_id. Course(course_code, instrument, grade_level, teacher_id). Primary key course_code; teacher_id is a foreign key to Teacher. (p4) Enrolment(student_id, course_code). Primary key (student_id, course_code). (p2) student_id is a foreign key to Student and course_code is a foreign key to Course. (p3)

Question 11

A school trains a model to flag students who may need extra reading support. The labels in the training data record which students teachers referred for support over the last ten years. In those years, teachers referred boys much more often than girls who had similar reading scores. On the records from those ten years, the model agrees with the teachers' decisions 92% of the time.

(a) Explain how bias has entered this model. [2]

(b) Explain why the 92% accuracy does not show that the model is fair. [2]

[4]

p1 [1]
(a) Says the bias comes from the labels, because the training data records the teachers' past referral decisions, which treated boys and girls with similar scores differently.
p2 [1]
(a) Says the model learns that pattern and repeats it, so it is more likely to flag boys and to miss girls with the same reading scores.
p3 [1]
(b) Says accuracy only measures how often the model matches the past labels, so if the labels are biased, a model that copies the bias scores highly.
p4 [1]
(b) Says a single overall figure does not show how errors are spread across groups, for example girls who need support could be missed far more often than boys.

Also accept

For p1, "historical bias in the data" earns the mark if linked to the past referrals.
For p4, "you would need to check accuracy or missed cases separately for boys and girls" earns the mark.

No marks for

For (a), "the programmers were biased" with no link to the data or labels. Nothing in the description says so.
For (b), "92% is not 100%". A perfect match with biased labels would be even more biased.
Saying the model is unfair because it uses a computer.

A full-marks answer

(a) The labels are the teachers' past referral decisions, and those decisions referred boys more often than girls with similar reading scores. The model learns from these labels, so it learns the same pattern and goes on flagging boys more readily and missing girls with the same scores.

(b) Accuracy only says how often the model agrees with the past labels. Because the labels are biased, a model that copies the bias gets a high score. Also, one overall figure hides how the mistakes fall: girls who need support could be missed much more often than boys, and that has to be checked group by group.

Section B: practice scenario

Section B is in the case-study format, on a practice scenario written by Spine. It is not the IB's case study for your exam year. Read the scenario first. Each question refers to it.

Practice scenario (written by Spine, not an IB case study):

Larkford is a small city with a community bike-share called LarkBikes. It has 40 docking stations, each with between 10 and 20 docks. Each dock has a sensor that detects whether a bike is locked in it and reads the bike's ID tag. Every station sends its sensor readings to a central server over the mobile phone network every 5 minutes.

Riders register in the LarkBikes app with their name, email address, home postcode and date of birth. When a rider unlocks a bike, a trip starts. The server keeps riders, bikes, stations and trips in a relational database. Each trip record holds the rider, the bike, the start station, the end station, and the start and end times. The database now holds two years of trips.

Every evening, a machine learning model is trained on the trip records together with the hour of the day, the day of the week, whether it is a school holiday, and the weather. For each station, it predicts how many bikes will be taken in each hour of the next day. A van crew uses these predictions to move bikes to stations that are expected to run out.

The 28 stations in the city centre are busy. The 12 stations in the outer districts were added last year and record far fewer trips.

Question 12

The manager of LarkBikes suggests a change. The model would also take, as a feature, each station's current number of bikes from the latest sensor readings. It would still be retrained once each evening, on the past trips and the stored sensor readings. During the day, the trained model would be run again every 5 minutes on the newest readings, to update the predictions for the rest of the day. All of this would run on the existing central server.

Evaluate this suggestion. [4]

p1 [1]
Gives a benefit, for example the current bike count shows what is happening today, such as a station emptying faster than usual, so the updated predictions let the van reach it before it runs out.
p2 [1]
Explains the processing cost in terms of training and inference, for example the heavy work (training) still happens once a night, and running an already trained model for 40 stations every 5 minutes is light inference, so the existing central server can cope without extra hardware such as GPUs or cloud computing.
p3 [1]
Gives a limitation, for example the van crew cannot change its route every 5 minutes, so it is the bottleneck, or a station that loses its mobile connection gives a missing or out-of-date bike count, so its updated prediction could be wrong.
p4 [1]
Reaches a supported judgement, for example adopt it, because the extra cost is small, but send the van an updated plan only every 30 to 60 minutes, and fall back to the evening prediction for any station whose readings are missing.

Also accept

For p2, "every evening's training now has more data to learn from (the stored readings), so it takes longer" earns the mark if it is still linked to training being the heavier job.
For p4, a judgement that the change is not worth it yet, because one van cannot act on the extra updates, earns the mark if it is supported by that reason.

No marks for

Saying the sensors would have to send data more often. They already send readings every 5 minutes.
Saying the model would need retraining every 5 minutes. The suggestion only runs the trained model more often.
Listing benefits only, with no cost or limitation, or giving a judgement such as "it is a good idea" with no reason.

A full-marks answer

The current bike count tells the model what is happening today, not only what usually happens. If a station is emptying faster than usual, for example because of an event nearby, the updated predictions show it, and the van can get there before the station runs out. The processing cost is small. Training, the heavy part, still happens once each night. Running the trained model for 40 stations every 5 minutes is inference, which needs far less processing, so the existing central server can do it without GPUs or cloud computing. There are limits. One van crew cannot change its route every 5 minutes, so the van, not the model, is the bottleneck. A station that loses its mobile connection would give a missing or old bike count, and its updated prediction could be wrong. On balance I would adopt the change, because it costs little, but send the van crew a new plan only every 30 to 60 minutes, and use the evening prediction for any station whose readings are missing.

Question 13

(a) State whether the LarkBikes model performs classification or regression, and give a reason. [2]

(b) Identify one feature the model uses and the label it learns to predict. [2]

[4]

p1 [1]
(a) States that the model performs regression.
p2 [1]
(a) Gives the reason that it predicts a number (how many bikes will be taken), not one of a fixed set of categories.
p3 [1]
(b) Gives one feature named in the scenario, such as the hour of the day, the day of the week, whether it is a school holiday, the weather, or the station.
p4 [1]
(b) Gives the label as the number of bikes actually taken from a station in a given hour, counted from the trip records.

Also accept

For p2, "the output is a quantity or a count" earns the mark.
For p3, "the temperature" or "whether it is raining" earns the mark as part of the weather.
For p4, "the number of trips starting at that station in that hour" earns the mark.

No marks for

For (a), "classification, because it sorts stations into ones that will run out and ones that will not". The model outputs a number of bikes; the van crew makes that decision.
For (b), "the number of bikes taken" given as the feature. It is what the model predicts, so it is the label.
For (b), "the rider's name" as a feature. The model predicts demand per station, and a name tells it nothing about demand.

A full-marks answer

(a) Regression, because the model outputs a number, how many bikes will be taken from a station in an hour, rather than choosing a category.

(b) One feature is the hour of the day. The label is the number of bikes actually taken from that station in that hour, counted from the trip records.

Question 14

The model predicts few trips at the outer-district stations, so the van rarely brings bikes there. Riders in the outer districts often find those stations empty.

Discuss whether it is fair to use the model's predictions to decide where the van takes bikes. [4]

p1 [1]
Explains where the low predictions come from, for example the outer stations have only one year of records and far fewer trips, so the model has little data and predicts low demand there.
p2 [1]
Explains the feedback loop, for example the trip records only count bikes actually taken, not riders who found a station empty, so empty stations record few trips, the model predicts even lower demand, and the van brings even fewer bikes.
p3 [1]
Identifies who is harmed and how, for example riders in the outer districts get a worse service for the same scheme, even if the model is accurate overall, and they may have fewer other ways to travel.
p4 [1]
Reaches a supported judgement that includes a change, for example record failed attempts to find a bike as unmet demand, or keep a minimum number of bikes at every station, and keep using the model above that minimum because it helps the busy centre.

Also accept

For p3, "the model can be accurate and still unfair, because accuracy is measured on the trips that happened" earns the mark.
For p4, a judgement that using the model is fair for now because the outer stations really are quieter, supported by a plan to check it once more data exists, earns the mark.

No marks for

Saying the model is unfair "because it is a computer" or "because AI is biased" with no link to the data.
Saying the outer-district riders should just walk to a city-centre station.
A one-sided answer with no judgement.

A full-marks answer

The low predictions come from the data. The outer stations were added only last year and record far fewer trips, so the model has little to learn from and predicts low demand there. This can feed on itself. The trip records only count bikes that were actually taken, not riders who arrived at an empty station. If the van rarely brings bikes, few trips are recorded, the model predicts even lower demand, and the van brings even fewer. The people harmed are riders in the outer districts, who are part of the same scheme but get a worse service, and who may have fewer buses or other options. The model could still be accurate overall, because accuracy is measured on the trips that happened. On the other hand, the outer stations may truly be quieter, and moving bikes to where they are used helps more riders in total. My judgement is that using the model on its own is not fair. LarkBikes should keep a minimum number of bikes at every station, record each time the app finds no bike nearby as unmet demand, and use the model to place the remaining bikes.

Back to the paper