JMP platforms and statistical methods

Distribution:
Histograms, box plots, summary statistics, tolerance interval

Graph Builder:
Scatter plots, simple linear regression

Multivariate:
Correlation, Scatter plot matrix

Fit Y by X:
One-variable linear regression

Fit Model:
Multivariable linear regression, Profiler

Objective

Build a multiple linear regression model to describe the relationship between the strength of concrete and the amount of various ingredients. The resulting model will be used to simulate the strength of a large amount of concrete made from a specific blend of the ingredients.

Problem statement

Concrete is an essential material used in almost all types of construction. The ability to form concrete into myriad shapes and sizes with different levels of durability makes it an ideal material to use for many different applications, such as a simple foundation for a small house all the way to the construction of bridges and large city buildings. Each application requires that the concrete used meets critical mechanical requirements, and in some cases, needing it to support massive amount of weight and sustain heavy vibration, such as that from an earthquake.

Concrete manufacturers perform frequent testing to ensure that each formulation they are creating for a specific application meets critical strength requirements. One such test – the concrete compressive strength test – applies a gradual increasing load onto a concrete cylinder to determine the maximum compressive stress the cylinder can sustain before fracture.

Concrete is comprised of four primary ingredients (cement, fine and coarse aggregates, and water) along with the addition of other materials, such as fly ash and blast furnace slag, which are designed to improve its mechanical and chemical properties.

To improve upon the understanding of how these different ingredients interact and influence the concrete’s strength, a data set was compiled. This data set consists of test results from a sample of 122 unique batches of concrete that had been made from many different manufacturers across the U.S. The amount of cement, coarse and fine aggregate, water, fly ash, and blast furnace slag used in each batch was provided. The data values provided by these manufacturers is the average maximum compressive stress across a set of cylinders tested.