Justifying CapEx Using Data Science…

You as the Chief Information Officer of a global multinational company are being pitched a new and improved IT system that promises to improve worker productivity by at least 10%. If that claim by the vendor was to pan out, your company based on data from HR stands to save tens of millions of dollars each year. But embarking on changing an entrenched system is no small feat. Employees would need to get retrained and that means an initial hit on worker productivity but then you stand to gain big in the long run. So how do you justify this upfront capital expenditure of acquisition, installation and training for this new system? You collect data. You run a pilot program by implementing this new system in a few locations say for a defined period of time and you record productivity gains or losses along with the relevant predictor variables and compare that to locations with the existing system. If your analysis proves the claims of the vendor, justifying that CapEx is going to be a cakewalk. If not, you just saved yourself all that CapEx and the associated heartburn.

So now that you have the data on this two scenarios, you want answers to these questions –

  1. Does the new IT system improve worker productivity?
  2. If it does, by how much on average?
  3. And is the gain enough to justify the associated CapEx?

Dataset – IT-System-Upgrade

Download instructions – Right click the link to the dataset and click¬†“Save link as…” to download the file. Then change the extension to csv from txt. Your Jupyter notebook has to be in the same folder as the downloaded dataset if you want to play around with it.

[Some outputs could appear to be a bit messy and difficult to read and if that is the case on the interface of your choice, follow the link to Github at the bottom of this pane. Things are much cleaner there.]

Image credit – John Spencer, Flickr