A university lecturer fed-up with his bus always being late spent months researching time-tables to prove to transport chiefs how inaccurate they were.
Ben Hanson, 31, decided to study the 'live tracker' displays on electronic noticeboards for two bus routes in his home city which he regularly used.
And he noted that when the buses were due to arrive in five minutes, they were late eight out of ten times.
Ben also discovered that buses could take up to a third longer to arrive at their stops than was predicted by the noticeboards.
The Leeds University physics lecturer has now submitted his 3,000-word research document to local transport bosses to help them make their system as 'accurate as possible.'
The frustrated commuter admits he was motivated by a 'personal grievance' that the electronic noticeboards which are supposed to be accurate were not.
And he used a computer programme to precisely measure - minute by minute - if busses were late on two routes in Leeds, West Yorks.
Ben now hopes West Yorkshire Combined Authority will use artificial intelligence and machine learning to make the electronic schedules more accurate.
He said: “It started off as a personal grievance, as an annoyance more than anything.
"I was just bored of waiting around all the time and realised I had the skills to look into it – so I did.
“I found that if I’m at a bus stop and it says five minutes on the board, 80 per cent of those buses will not be there within five minutes - they will take longer.
“And the solution I’ve suggested is machine learning and artificial intelligence.”
Ben, who lives between the villages of Farsley and Stanningley, Leeds, West Yorks., said he carried out his investigation during two-week periods in May and August this year.
The boffin collected his data over the internet, via a computer programme, which precisely compared bus arrival times with electronic notice board predictions.
He said: “The goal was to get onto the internet and find out where the live timetables are getting their bus data from - which you can actually do quite easily.
“Once I did that, I created an automated way of doing that by writing a computer programme that checks for you.
“And I made that computer programme pull down every single minute for two solid weeks."
Ben’s impressive research examined the number 72 bus, running between Leeds and Bradford, along with the number six bus, which travels through the city centre.
And his analysis revealed that the "live tracker" boards on the routes were "systematically incorrect" - as buses arrived up to 33 per cent later than they predicted.
He said: "The main thing I would want to point out is the times on the boards are not real times. It’s just their best guess – and their best guess is very bad.
"If it says six minutes on the board, you would expect it to change to five minutes after one minute, but that’s not true – that’s not what happens.
“If you actually measure it, it says those times for sometimes even three minutes. So whatever their prediction mechanism is, it’s just wrong."
Ben said he conducted his research on a mixture of different days to ensure that his data wasn’t influenced by rush hour traffic or one-off events.
He explained: “The research showed you can’t blame it on time of day or area, because I crunched all the numbers so that each time of day and type of day was separate.
“It happens independently of that, so there was something else going on."
And Ben said after he submitted his research to the local council, he’d received positive feedback and had been invited to talk with them about it.
He said: “The response has been quite nice. I expected them to be quite angry and start denying everything. But they didn’t.
“They said they were quite happy to work with me or whoever is the expert at solving this issue.
"They even invited me to a chat on Wednesday about how we might go about solving it.”
Ben said the problem he had found could be solved using artificial intelligence and machine learning, which would more accurately estimate bus arrival times.
But despite his success, he said he didn’t want to ditch his day job to help other councils improve their electronic timetables boards.
He said: “If it turns out there’s national interest, rather than just a local one, then it could be the solution I’ve suggested is machine learning and artificial intelligence.
"That scares people, but it’s actually a mechanism by which a computer programme can learn using historical data how long a bus is going to take and update its predictions going forward.
"But my job is a physics lecturer and researcher. I did this out of interest. It’s not something I can do, and I’m not going to quit my job to do this.”
A West Yorkshire Combined Authority spokesperson said: “Our real-time service tracks hundreds of buses approaching 14,000 bus stops in West Yorkshire, as well as services into South Yorkshire, North Yorkshire and other neighbouring counties.
“Real-time journey predictions are displayed on over 2,500 on-street and bus station displays, and over 2 million system journey prediction messages are generated every day.
“We and our partners recognise the importance of reliability and accurate information for people who use the bus.
“We continuously monitor its accuracy and update both the hardware and software to refine the arrival time predictions.
“We’d like to thank Ben for the summary of his analysis into these two stops, and hope that he can share his research with us to help us ensure the system is as accurate as we can get it."