# Thread: Logic to Business problem

1. ## Logic to Business problem

Hi,

In my company there are 15000 employees currently out of which 3000 are engineers with different years of experience.
Over the last 10 years the company has approximately 40000 employees out of which there are around 6000 engineers who joined and then resigned during this period.

My present HRMS application has unique entry for all these 40000 employees along with their salary details.
The salary details of the 3000 engineers are also included here. However their salary range is not fixed. It varies on their experience, trade, soft skills etc.
Over the 10 years the monthly salary range for engineers vary from $8000 -$ 40000. The salary range for them in last 5 years is between $27000 -$ 40000

I am designing a new Recruitment application that needs to have the average salary for each job title.

So here's the confusion. If I need to know the current average salary of engineers, what will be the most accurate logic? A direct average salary of the 3000 engineers will not help due to the high variation.

Thanks in advance !!!!

2. ## Re: Logic to Business problem

Hey sarathpillai.

Do you have any data you can use to get a distribution of the relevant engineers?

If you have a bunch of data for all engineers then you can use a subset of that data (also known in statistics as a conditional random variable) and then either calculate the median or the mean of that distribution.

Alternatively if you want an interval, then you can get a confidence interval (typically 95% is used) and report that.

In terms of the tools to calculate things like this, you can use anything from Excel to statistical packages like R or SAS.

3. ## Re: Logic to Business problem

Thanks Chiro,

Can you please enlighten me with a simple example?

4. ## Re: Logic to Business problem

How do their salaries vary over time? For example do they increase by $3800 every year for 5 years then increase by$2600 every year for the next 5 years?
Do you have access to the individual 3000 present salaries? If so it would help to calculate the sample variance with the formula

$\sum_{i=1}^{3000}\frac{(x_i-\bar{x})^2}{2999}$
Where (x1,x2,x3, ... xi, ... ,x3000) are the 3000 salaries and $\bar{x}$ is the sample average salary.

5. ## Re: Logic to Business problem

I think your problem is in assuming that "engineer" is a job title. It's not. You need to be much more specific in creating pools of salary data by expertise and experience. The market for electrical engineers drives an entrely different salary structure than for civil engineers or mechanical engineers, so you need to differentiate by field of expertise. Also in engineering there are various levels of capability - you should distinguish between entry level engineers fresh out of school versus project engineers who can work more autonomously versus senior engineers who lead projects. I can say from experience that developing a consistent set of job titles and salaray metrics for all engineers in a high-tech company is not easy, and if you try to lump them all together you will end up with inconsistent data and not terribly useful results.

6. ## Re: Logic to Business problem

I do agree with you Ebaines.
'Engineer' is just for example. We do have specialized designation like Electrical, Mechanical etc. But we do not have records captured for their niche skills, soft skills etc based on which the salary variation happens.
I know that we would be having inconsistent data, so i am looking for a logic which would minimize the severity of its effect... thanks !