Developing Data Products Course Project

This application is part of Developing Data Products Coursera course project http://www.coursera.org/learn/data-products/.

The application predicts the class of a car with random forest model based on engine displacement, number of cylinders, city mileage and highway mileage. The model was trained on mpg dataset from ggplot2 package. http://ggplot2.tidyverse.org/reference/mpg.html

This presentation is R presentation created in Rstudio.

The shiny app pitched by this presentation is at:http://georgymakarov.shinyapps.io/ddp_course_project/

The source code of the app is at: http://github.com/GeorgyMakarov/Shiny-car-predictor

Model uses random forest algorithm with cross-validation.

control <- trainControl(method = "cv", number = 5)
set.seed(seed)
fit.rf <- train(class ~., data = training, method = "rf", metric = "Accuracy", trControl = control)
pred.rf <- predict(fit.rf, training)
conf_m <- confusionMatrix(pred.rf, training$class)
conf_m$overall["Accuracy"]

 Accuracy 
0.8829787