Steel Strength Prediction

Steel is widely used for a variety of applications. Steel consists of a number of chemical components such as Iron, Chrome, nickel, etc. These components gives it strenght and a host of other properties

Stength of steel is one of the most important characteristics that has a bearing on the life and costs for a number of products and structures that are built using steel.

With an objective of predicting the strength of steel based on eight parameters, we have built a neural network and tried predicting the strenght of steel using the neural network model.

We have explored the single layer and multi layer neural network for predicting the strength of steel and evaluated the performance of the model based on the correlation of the obtained values with those of the actual values

## 'data.frame':    1030 obs. of  9 variables:
##  $ iron      : num  298 318 363 190 305 ...
##  $ chromium  : num  0 0 189 0 204 ...
##  $ nickel    : num  107 126 0 125 0 ...
##  $ vanadian  : num  186 210 165 162 204 ...
##  $ manganese : num  6.1 6 11.6 9.9 0 11.6 10.1 4.6 10.6 0 ...
##  $ coarseness: num  879 861 945 1088 965 ...
##  $ fineness  : num  815 737 756 803 631 ...
##  $ age       : int  28 28 56 14 7 7 91 3 28 7 ...
##  $ strength  : num  42.6 40.1 77.3 22.7 25.9 ...

Normalizing the values of the variables in the concrete dataset and evaluating the obtained values with original values

Summary of values form the normalized dataset

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##  0.0000  0.2664  0.4001  0.4172  0.5457  1.0000

Summary of values form the original dataset

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    2.33   23.71   34.45   35.82   46.13   82.60

Single layer neural network model

Co-relation between actual and predicted values for single layer neural network is

##            [,1]
## [1,] 0.79746755

Multi layer neural network model

Co-relation between actual and predicted values for multi layer neural network is

##              [,1]
## [1,] 0.9232240214

As can be seen the performance of the model with multi layer neural network is better than that of the single layer neural network