PMS 3X

3X PMS is a Perfomance Measurmenet System project created by 6 students completing their Masters in Data Science at the Bologna Business School

• The topic for 3X PMS is Cost driver detection and prediction.
• The main business problem that is being solved here is how can we predict the Total cost helping the customer to determine whether he should accept the proposition from a client or not

The project members are

• Mohammed Topiwalla
• Sanchita Kumari
• Luca Pedretti
• Vallerio Trotta
• Marco Lantermo
• Guilia Capestro

• Powerpoint for the project is here
• Regression output for the project is here

PMS Analysis

Note: 3X PMS project uses only R 3.4.X and Tableau 10.X and above for analysis and prediction

Installation

install.packages("shinydashboard")
install.packages("Boruta")
install.packages("data.table")
install.packages("ggplot2")

Approach to solve the problem

• Understanding the data, the tableau result is here
• Cleaning and preparing the data
• • Some of the cleaning was first done manually in excel - for instance removing rows with negative numbers
• • The pending cleaning was done in R. The code is here
• • Note: The code that loads all libraries is here
• Next step was to find the cost drivers, for which multiple techniques were used
• • Boruta (Random forest featue selection technique). The reults of boruta are here
• • Checking correletarion among major cost drivers
• • Boruta on unit costs . The results of boruta are here
• • To find the most imporatnt cost drivers we ran a few regressions. The code sits here. The results are here
• • Finally after running few other regressions, we understood our cost driver. The output is here
• Once we undertsood the cost drivers the next task was to be able to predict the total cost based on a few entered parameters.
• • To try it out go to https://mohammedtopiwalla.shinyapps.io/PMS_Analysis/
• • upload the data from here
• • Enter the parameters you like and change them dynamically to see the reults change.
• • The code for Rshiny is here