Statistical Modelling

Statistical modelling

The essentials

Statistical modeling is a mathematical representation of reality, essentially an equation which can generate results reflecting what goes on in real life.

Your data is based on what you have observed or recorded, under the conditions that were present at the time. But what if you wanted to see how those results would change under different conditions? What if demand suddenly rose? What if there were less staff? What if you raised prices? How would these changes to the environment affect your outcomes?

Statistical models describe the relationships between the variables in your data. Using these relationships, the model then attempts to simulate the process which generated the data; it attempts to simulate reality

What can modelling tell you?

Understand why you get the results you do – what are the biggest factors? How are they linked? What isn’t important.
Predict future behaviour and forecast outcomes – how will we perform next year? What will our output be for the next 6 months?
Investigate “what if” scenario’s such as increased demand, increasing prices.
Assessment of risky outcomes and therefore risk management.

Modelling, even for seemingly simple situations may not be straight forward and it is always important to involve an experienced statistician early if you are thinking of developing a model.

If you want to know where your data has come from and why, we can construct a model that will help you to understand and appreciate it. Our consultants are all highly experienced and internationally respected when it comes to statistical modelling.

Types

The list below is by no means exhaustive, but highlights some of the modeling techniques that are available:

Regression models (including linear, generalised linear and logistic models)
Non-parametric models
Semi-parametric models
Fixed, random and mixed effects models
Bayesian and graphical models
Time Series models
Survival models (Kaplan-Meier)
Classification and regression trees