Help - Expert Statistician required

[Panda]

1st a bit of Background. I can't go into too much detail as there is huge commercial implications to the project, but I'll outline what were trying to do in general terms. And please remember I'm a rocket scientist not a mathematician so please go easy on me.

My department hypothesised that when designing vehicles to operate in hostile environments, rather then using specially designed components and subsystems for the whole of a vehicle some could be every day off the shelf stuff which didn't even need modification. We did some tests and showed that you could apply rules of thumb on how certain items would survive.

We have since been doing more exhaustive testing where we take a component or subsystem, anything from a solenoid pack from a toy car to a network hub, which does a similar job to the specially designed units in our vehicles and subject them to a well established test that replicates a specific environment, and then see how they fail.

This is where the problems start. The test set up means that whilst there are a dozen parameters that may affect failure, we can only control the nominal intensity of test and the number of events in the test.
We know that two identical components will also vary according to a number of parameters most of which we can't control.

So the data we get out for each component is the number of events it experienced at a nominal intensity before a fault condition occurred. If the fault can be rectified we will continue testing, otherwise we get a new unit.

We then want calculate the statistical likelihood that during a mission life where N events will be experienced by the device it is likely to survive or it will probably fail. The implications being that we can send 5 vehicles costing £500K each rather than one costing £50M and get the same probability of success.

When we try and analyse the data because every test is different from every other test we end up with massive uncertainties in the results. Analysis indicates that failures are randomly distributed according to event intensity rather than cumulative and life stress analysis does produce results that roughly fit our data.

The key is, that if this is to work we need to do a small number of tests as the time involved in testing components can easily remove any cost benefit over using a specialist device.
The data we get is a pass/fail after exposure to a number of event of given intensity, rather than the exact intensity required to cause failure.

Apart from employing a full time Statistician, What do we need to do to determine what the individual component reliability is likely to be based on a small number of tests with poor control over parameters?

 

[matt grime]

My consultancy fee is 100GBP per hour for this kind of work, as should anybody else's be, if not more. If there are indeed huge commercial benefits then you won't have any problems coughing up the cash.

 

[Panda]

My consultancy fee is 100GBP per hour for this kind of work, as should anybody else's be, if not more. If there are indeed huge commercial benefits then you won't have any problems coughing up the cash.

Not the most helpful answer. I am looking for techniques and pointers not consultants who don't have an understanding of the underlying physics. Once we have a technique that seems to work then we can look at investing in software and other tools to analyse the data.

 

[cristo]

People give out their advice for free on this forum all the time in order to help people with their homework, or to help them in their studies. However, this is a completely different situation. You are asking for someone to give you their advice, for free, on a project that you think will have huge commercial implications in the future, i.e. will make you lots of money. I don't think you will get many qualified people helping you for free on designing a commercial product! For example, if you went to a university's maths department and asked them to help you, do you think someone would do it without charge?

I agree with matt: if you want a professional mathematician's help, then you will have to pay his fee!

 

[matt grime]

Not the most helpful answer.

Oh, so sorry.

I am looking for techniques and pointers not consultants

you are looking to get free advice that will help your company make money. Quite frankly you can sod off.

who don't have an understanding of the underlying physics.

that's rather insulting - you have no idea what physics I do or do not understand

Once we have a technique that seems to work then we can look at investing in software and other tools to analyse the data.

once you have ripped off free advice telling you what you need to know you will put the information to use in making you money, you mean.

You want statistical consultancy (and you ask for an expert statistician), so go pay for it, and stop trying to get freebie advice from people's good will. I think you'll find that the firgure I quoted you is not unrepresentative of the amount of money you will need to pay in order to get the expert advice you crave for commercial purposes. Certainly you won't get less than 500GBP per day from the statistical consultancy I'm familiar with.

The university of the west of England, (UWE) does statistical consultancy work. Why don't you go ask them?

 

[Panda]

I am not asking for product design, we are capable of doing that ourselves. I am asking for advice on how to analyse data. The data analysis does not add any value to the product it just helps me understand the test results better.
If it is easy then say it is easy. If it is difficult and requires considerable time to even assess the problem then say so. If I really like the advice I might even favour your academic institute for tackling the wider problem. I am not prepared to give somebody £10k just so they can tell me the problem is too hard.

It is a sad indictment of certain academic institutes in the UK that they are not prepared to even say if they are capable of answering a question without being paid. I have paid out over £50k to universities so far this year just to get academics to tell me the problem was not within their capabilities, and this is putting a dent in the funds available to those who are capable of doing the research.

Compare that to the French, German and Italian institutes we deal with, who always give assessments free of charge. It costs me less to fly to Rome to speak to their physics department than going to the likes of the Bristol University Maths department, and if it’s a choice between two days in Rome or an afternoon in Filton I’ll take Rome any day. That is why UK universities have lost out on over £500k’s worth of follow on research over the past 12 months and my Air Miles account is looking very healthy.

 

[matt grime]

Put together a proposal, and go to UWE, they'll tell you if they can do it. As it is any A-level student should be able to tell you what you want to know, so I'm reasonably confident that UWE can answer your questions.

 

[Panda]

It saddens me that you show such little attention to people’s questions. If this problem was "A-Level" standard then one of the graduates looking at the data would have solved it by now.

I may well offer the task to UWE, one thing that is for sure though is that based on this example of your professional approach it won't go to the University of Bristol Mathematics department. You can say that your views do not represent those of your employer but your attitude still reflects on them.

It is always foolish to jump in with groundless preconceptions about somebody and anti-commerce biases before establishing how many opportunities you are shutting down.

It is a good job that Bob Evans is not as closed minded as you are when it comes to finding opportunities for the University. You should take care before rubbishing people’s questions, not everybody on this forum is an A-Level student.

 

[matt grime]

It saddens me that you show such little attention to people’s questions. If this problem was "A-Level" standard then one of the graduates looking at the data would have solved it by now.

I'm going by precisely what you wrote in the first post. Perhaps a-levels are easier these days, or perhaps your graduates aren't that good. Anyone with a reasonable knowledge of stats can answer your question. I'm going to refuse to because you clearly want free advice for a commercial project.

There is also a difference from having once learned something some years ago and being able to remember what it was.

I may well offer the task to UWE, one thing that is for sure though is that based on this example of your professional approach it won't go to the University of Bristol Mathematics department. You can say that your views do not represent those of your employer but your attitude still reflects on them.

It is always foolish to jump in with groundless preconceptions about somebody and anti-commerce biases before establishing how many opportunities you are shutting down.

It is a good job that Bob Evans is not as closed minded as you are when it comes to finding opportunities for the University. You should take care before rubbishing people’s questions, not everybody on this forum is an A-Level student.

I'm closed minded? You're the person ignoring the disclaimer - I in no way speak for the university of bristol. Look beyond the personal and you might get somewhere yourself.

You think it inappropriate of me to have given you short shrift. I think it highly inappropriate of your company to attempt to solicit advice for a project with 'large commercial implications' on a website, using a pseudonym, and without any indication that you would give credit in any form to those who would help you.

Had you actually approached an expert statistician I'm sure you'd have found many willing people - there is nothing a mathematician likes more than explaining maths, especially to someone who knows less about the subject than them (Tom Koerner).

 

[CRGreathouse]

Wait, wait. Matt Grimes' suggestion seemed very reasonable to me -- rather than hire a full-time statistician, you can just pay someone to consult with you to set up the basic system and give you advice. While a full-time person would cost a bare minimum of ~40,000 GBP, 10 hours of consultancy could be had for a digit or two less.

For Matt's time 100 GBP / hour is probably fair. If you go with someone less experienced you could surely get a better rate.

 

[Hurkyl]

Panda: If I had to guess just based on the information in this thread, it sounds like you want conclusive results from a small amount of inferior data -- and the academics have rightly told you that it can't be done.

A staff statistician (or consultant) might be able to work with your engineers to design an experimental setup that gives better results, and would be able to estimate how much data you need to get results.

But whatever you wind up doing, you have to be willing to accept that the right answer might be that it's not cost effective to do the research into off-the-shelf components. (Of course, that may or may not change if you can amortize this cost over the next five projects)


That said, I'm not entirely sure this is an appropriate forum for your inquiry, and I am entirely sure that I don't like the tone of this discussion. matt's response may have been flippant, but certainly not out of line, and I won't tolerate your threats here. Thread closed.