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

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

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

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

mmd52 / pms_analysis Goto Github PK