Shiny Product Illustration

Medhat Farag
July 16, 2018

Part of Data Science Coursework @JHU on Coursera

Product General Idea

The presented data product represent a simple prediction apllication based on IRIS dataset

  • The application starts by showing a pair plot between all variables of IRIS dataset
ggpairs(data = iris, columns = 1:4, aes(colour = Species))

plot of chunk unnamed-chunk-1

Procedure Illustration

  • The user has to enter 4 values of Sepal and Petal length & width respectively. 4 input fields are assigned for those values in the UI. The lower and higher limits of the input fields are the ranges of these values in IRIS dataset

  • The 4 values are used to construct a data frame to apply the prediction model on it later

  • Boosting algorithm is used to predict the type of flower corresponding to the supplied dimensions

  • The algorithm is trained on the whole 150 observations of the dataset

mdl <- train(Species ~ ., method = "gbm", data = iris)

Procedure Illustration (Cont.)

  • The Boosting prediction algorithm is applied to the selected point data frame to get the class of flower correponding to the entered dimensions

  • The output class is colorly coded according the the colors used in “ggpairs” plot (Red: Setosa, Green: Versicolor & Blue: Virginica)

  • The interface of the input fields & predicted result are displayed in 3 columns format as shown in the below snapshot

alt text