Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I would like to calculate the value of bacteria on 4 surfaces $i=\{1..4\}$. A person touches some of those 4 surfaces at random and a count is made on their finger after each surface contact ($x_i$).

Someone lost the bacteria count ($x_i$) after each surface but I do know the total count (X) on a persons's finger after they've touched a number of surfaces. I also know which ones and in which order.

What I know:

  1. Final bacteria count on a person's finger: $X$
  2. Transfer efficiency from surface to finger: $PT_i=\displaystyle \frac{\text{Finger contact area}}{\text{Area of surface}_i}\frac{1}{\gamma_i}$ where $\gamma$ is a surface dependent constant.
  3. The number of times the person touched a particular surface: $h_i$.

If I had surface counts $C_i$, the summation of bacteria is linear: ie $\begin{eqnarray} h_1C_1PT_1&=&x_1\\ h_2C_2PT_2&=&x_2\\ \vdots\quad &=& \quad \vdots\\ h_iC_iPT_i&=&x_i \end{eqnarray}$

such that summing over all surfaces $i$ the total count x is: $\displaystyle \sum_i h_iC_iPT_i=\sum_i x_i=X$.

Can I back calculate $C_i$, without $x_i$ even statistically or probabilistically?
Best regards.

share|improve this question
Did each subject touch all 4 surfaces, or did some touch some surfaces more than once? If the latter, simultaneous equations spring to mind. – Michelle Jan 25 '12 at 19:12
Does the finger contact area in your expression for $PT_i$ vary with each touch? – jbowman Jan 26 '12 at 1:04
@Michelle The subject(s) don't necessarily touch all four but any surface can be touched a number of times. – HCAI Jan 26 '12 at 8:12
@jbowman It could do. – HCAI Jan 26 '12 at 8:13
Do you have a measure of length of touch for each trial? – Michelle Jan 26 '12 at 9:50
up vote 2 down vote accepted

I've only dipped the top of my toe into operations research, so I welcome comments on this suggestion. Basically, you have what I think is an optimization problem (trying to get the most accurate estimates for each surface). I think you have a problem that can be solved through solving as a series of simultaneous equations using either linear or nonlinear programming. This falls squarely into operations research, and is outside my area of expertise.

I don't think you can use a Markov Chain process for this as you don't have transitional probabilities.

share|improve this answer
Interesting. Could you advise as to what part of linear programming this might come under? I do actually have the transition probabilities but they're only a guess at best as I didn't observe the process. Would they help do you think? Thanks and best regards – HCAI Jan 26 '12 at 19:01
On reflection I don't have the transition probabilities between surfaces. Only the probability of touching surface $i$ – HCAI Jan 26 '12 at 19:19
Sorry, I can't be of more help with the LP/NLP as I have no experience or qualifications in that area. I'm hoping for an OR professional to chime in. – Michelle Jan 26 '12 at 19:20
Thanks for your input Michelle, it's appreciated. – HCAI Jan 28 '12 at 8:17

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.