Manual Data Collection

In recent years, corporate databases have improved, companies have invested in information systems, and data mining techniques are more commonly used. Still, we find that we are frequently required to use manual data collection techniques in support of our simulation work.

We have been working with and recommending our friends at Quetech since 2001 whenever we, or our clients, require assistance with time studies. Quetech is the developer of WorkStudy+. This software (for the Palm OS or Pocket PC) allows for streamlined and easy time studies using a handheld PDA.

Here's a quick run-down of how it works:

1. Develop your time study template by following an extremely user-friendly interface in Microsoft Excel.
2. Transfer the template to your handheld(s) using a USB cable.
3. Manually observe your process using a touch pen to record the start, end and category of tasks as well as many other customizable fields. Repeat until you've completed your data collection plan.
4. Transfer the data back to Excel for reporting and built-in statistical analysis.

We recently coached BJC HealthCare through a manual data collection effort with the goal of understanding the resource requirements to dispense medication within a hospital. The BJC study was covered tasks performed by the nursing staff within each unit, as well as order fulfillment tasks performed by pharmacists and technicians. WorkStudy+(combined with a few inexpensive PDAs from Walmart) proved invaluable in the manual data collection effort.

If you find yourself faced with a significant manual data collection project, consider this cost-effective, efficient solution. The days of stop watches, clipboards, and data entry will certainly not be missed.



Distribution Fitting

I recently came across a good whitepaper about distribution fitting from our friends at Vose Consulting (Vose offers Monte Carlo simulation software and consulting).  This paper provides a nice reminder about curve fitting as well as a detailed lesson in goodness of fit tests. This quick read will come in handy for improving the accuracy of your SIMUL8 models. The paper is available in PDF format:

Fitting distributions to data and why you are probably doing it wrong

by David Vose
Feb. 15, 2010

Data, Data Everywhere and Not An Expert in Sight

If you've built a few simulations, its a good bet that you've been through projects where you just didn't have access to all the data the textbooks said you should have. In such cases, the enterprising simulationist is usually left with no choice but to deal with this situation by interviewing appropriate subject matter experts (SMEs) and making what amount to expert guesses.

Click to read more ...

Modeling Variability

A recent project highlighted a different way of needing to look at variability.  The normal practise of gather, analysing and entering process times into SIMUL8 just did not work. The normal method would be:

1. Collect data into a spreadsheet.
2. Manipulate the data if necessary to get it into the form of minutes per piece.
3. Paste the data into your favorite statistical software to identify the distribution which best represents the variability of your system.
4. Enter the resulting distribution into SIMUL8 as a work center's operation time.

Click to read more ...

Gathering and Preparing a Data Set

Use the right tools. Data analysis software has come a long way in recent years. If you haven't researched what's available lately, you might be surprised at how much power you can get for relatively low cost. At a bare minimum, you should invest in a tool to help with distribution fitting. SIMUL8 has recently introduced the Stat:Fit plug-in. We think that ExpertFit is another excellent choice. Excel is good for some simple tabulation, but there is a wide array of techniques out there that are invaluable for exploratory data analysis. If you are working on projects involving data sets of any substantial size or complexity, you will appreciate the features of a complete statistical analysis package. There are many choices available on the market. For full-fledged data analyses, we use Statistica, from StatSoft, because it offers an impressive array of features at a very reasonable price. The right choice for you will depend upon your needs, background and budget.

Click to read more ...