Everyone, here in India, is familiar with Quick delivery platforms. There are different possible models (and systems based on them) for doing deliveries. With this project you can test how effective various delivery systems are by simulating customers, generating artificial orders, running them through multiple delivery systems, and analysing the delivery metrics hence collected.
I worked on implementing a truncated normal distribution sampler, alongside majorly replanning my solution.
Earlier I was thinking, I would model an order as a process but now I understand it’s better to model a real-world process as a “process” instead of modeling a non-active component (e.g- order) as a “process”.
I learnt about “resources” in simpy and am planning to implement delivery riders and order processors as a resources which the delivery process of a given order utilises (in order to? get delivered)
What did I work on?
I restructred project alonside refactoring the code for superfast iterability and modifications.
IDK refactoring and getting things in perfect structure feels so exciting!
Final Structure:
What did I work on?
I completed simulating the ordering system.
What it means?
I am now able to successfully create required number of customers and keep them alive simulatneously in my simulation. A customer can now successfully place an order with the dark store.
Effectilvely this means something like the DarkStore is receiving orders from modelled customers.
Now the next natural step is to model the Delivery systems which will be fulfilling the simulated orders.
What I struggled with?
I wasted a lot of time trying to come up with a perfect way to simulate order data (SKUs, Items, Prices, etc.) as realistically as possible. But - the fact was - changing the order data has almost no effect on delivery metrics which determine how effective a delivery system is. It only affects order preparation time.
What’s nxt?
I will be writing code to simulate the delivery systems.
What Did I work On?
I tried simulating a customer in the ordering process.
Along the way I realised that first I’ve to assign a unique location to each customer when I create it. So I spent the rest of the session trying to come up with a good enough model for location assignment and delivery route length determining (BTW I used a city grid model).
What did I learn?
I spent a lot of time(10X of logged) trying to learn SimPy to code the simulation.
Problems along the way?
There was a constant feeling of being stuck because I couldn’t find the right model for location assignment, right way of using SimPy.
Apart from that everything was effortful but pretty fun!
What’s Next:
I will complete the order simulation process which will allow me to generate orders from customers artificially for then feeding to respective delivery systems for testing.
