I started my journey of what eventually we would summarize as “Procurement Data Scientist”. Interestingly, I found some nuggets while pondering over my last 3 years notes on Procurement excellence roadmap!
These are my personal opinions as a procurement category manager, having done the granular “boots on the ground” work in procurement and logistics operations! What appeared as grunt work at that time has now become my forte to judge whether a technology product is a hype or has some potential!
Here are some notes that eventually might help you to chalk your own journey!
2015: Python and machine learning!
My Notes: I needed to learn how to code! Quite a daunting task considering I was in Procurement! But, the motivation came from the simplification it could brought to my day to day work!
We can easily solve my vendor award allocation for my commodity volumes! Linear programming will be the best solution! Took me a year to learn the basics and implement it.
2016: More machine learning and Procurement use cases:
Got introduced to Andrew Ng’s machine learning program on Coursera! Then I pondered over standford lessons on machine learning (free of cost) on YouTube! Repeated the entire course Atleast 3-4 times to get a grip of basics!
I strongly suggest you to spend the time it needs! Chatgpt is great but you wont get very far if you dont understand how to build stuff.
Finally understood how to implement simple regression model and maths behind it! Implemented it as work to do some time series forecasting on my commodity consumption patterns! Dealing with seasonality and achieving a white noise model was so difficult! Later this will be termed as demand forecasting).
Here is my most effective hack! Build a list of top 10 or 20 problems you want to solve. Dont worry about complexity or time! Rank them in order to execution difficulty (this can change over time, so it doesn’t matter even if you are wrong!) and secondly, mention their impact on your business operations (low, medium or high). Choose the easier one to execute as your first project!
2017: I got introduced to blockchain by consultants!
I struggled to understand any use case in procurement! Didn’t pay any attention! I Progressed to learn basic neural networks and maths behind it.No use cases in Procurement at that time!
2018: More blockchain hype and smart contracts pitches!
Again, I couldn’t see any meaningful execution value! A centralised vendor platform was far more effective in my view (and I still hold this view)
2019: I got obsessed over spend analysis!
So many use cases, so much value! Explored how to establish a good taxonomy and its importance! Hello, classification models! I experimented with using classification models to cluster similar items! Early success!
2020: More spend analysis and basic machine learning:
Spent more time in perfecting classification models! More importantly, I experimented with concepts of context based spend analysis and user journies. Meaning, moving beyond dashboards triggered as a mission because I quickly realised spend analysis industry was heading towards dashboard only product segment!
So, I wrote a book on spend analysis! Crafted what I saw my way to finding cost saving projects in vast spend data using only excel ! It’s the approach that matters more and not the use cases!Do you see my point here?
In summary, you need use cases to validate the technology and not the other way around. I’ll expand more on this in my next post!
Comments