It's always exciting when the data visualization or analysis you did is used to push forward a movement or a cause. Most would agree this is probably one of the greatest satisfaction we derive as data scientists (again, I'm using this title loosely). So, I got to know of this social initiative called SG Food Rescue. They were trying to tackle food waste in Singapore by collecting unsellable but edible food from vegetables/ fruits wholesale markets and then distributing them to those who need it i.e soup kitchens and charitable organisations that feed the needy. And of course, I spot an opportunity on the usefulness of data, visualizations and analysis in quantifying efforts and impact (occupational disease, really). I popped the idea to the co-founders and they were interested to explore how this could work. To better assess what type of data is necessary, they asked me to join them in a rescue session. Definitely, walking the ground is especially important and valuable; I signed up as a volunteer for two sessions and was pretty amazed how data science led me to rescuing veggies and fruits!
Being part of the action allows me to:
Have a better understanding of the work and appreciation of the cause (business understanding!)
Understand the challenges of data collection
Think of ways how process can be streamlined to facilitate data collection
Place more attention on certain areas during data cleaning
Cross-validate the data with what I see in reality
In this case, data collection is pretty taxing as it is a manual process where volunteers will have to weigh and impute the data (into Excel). However, to be able to state the impact of their efforts confidently and convincingly, they winged it and decided to track the results of their collections. While for data science, it's always the more data the merrier, we decided to make do with the data for eight sessions given the tedious data collection process. The data was, however, sufficient for us to have a high level overview on certain statistics. Around 1.5 tonnes of food was collected every session on average and certain food are more commonly overstocked such as capsicums, (sweet) potatoes, radish, zucchini and Japanese cucumbers. However, there isn't much consistencies in the types of food across sessions i.e. there is quite some variety in the food collected by day of the week/ month so beneficiaries can expect a different combination of food (which is a good thing!).
On a separate note, I got to learn a lot about the names of different veggies! For example, I never knew what snake gourds were. And the same veggie can be called different names by different persons (eg. scallions vs spring onions). Hence it's important to group them together for consistency during data cleaning. I also ended up spending much time trying to figure out the difference between zucchini and Japanese cucumbers.
With more data, we will be able to do other analysis such as market-basket analysis (whether certain veggies/ fruits tend to come together in a collection) or forecasting volume of veggies/ fruits. That would be useful for ideation on recipes maybe! Or rather, identify whether certain combinations are complements and it would be useful to re-look at the volume stocked.
Alternatively, the dashboard can be found here.