DoorDash Data Delivery

Food. Who doesn't love to talk about food?! It's one of life's greatest pleasures. We as humans LOVE food. We love to eat it, we love to drink it, we love to "vegan"-ize it. We plan for it, we prepare it, we savor it, and then we think about the next time we are gonna eat it. There's something sweet, smooth, salty, tangy, chewy, or crunchy we all enjoy. Food is the best! Right now you are probably salivating from all of this food talk. I'm sorry! (kind of).

Food and Data

I've recently joined a data bootcamp where I am looking to explore and expand my data skillset. And I thought it would be great to look into a topic where a lot of people have a common interest in, like food! So here we are. Talking about food and data. And the great thing about data is that you can really get as specific as you want with whatever you want. Among the various avenues I could have chosen to go down, I thought it would be interesting to explore the food delivery aspect of food data.

Over the last several years, the food delivery industry has grown with the rise of cellphones, especially during the COVID times. Many people have shifted into this new way of living. Instead of taking the time to prepare, cook and clean after your food, food delivery allows people more time to enjoy their food and do whatever else they want: work, watch tv, play with their kids, etc.

Now, I personally am an old-school guy at heart. I find pleasure in taking time in the kitchen to chop the lettuce, saute the onions, and slow cook the chicken. However, that does take a lot of planning and time involved. And I get that a lot of people don't live like that. And I'll be honest, every once in a blue moon I do use DoorDash® it is very nice and convenient during those times. So this project I've undergone is to analyze the various aspects of what food delivery is and does, and what the people/data are saying.

Why is this even relevant?

That's a good question. The reason this information is helpful is because it can give food delivery companies insight on how to improve their processes, generate more revenue and learn which products aren't selling. Instead of spending an equal amount of time with every item, they can target their focus to what's bringing them the most value.

Key Findings

So from the data that I have gone over and with help from Excel, it was interesting to see:

• A total of $1.24 million was spent on DoorDash® from the sample in 2018
• 75% of the spend variance can be explained by income levels
• An average of 183 people joined DoorDash® each month with November and December being the least and January being the greatest growth month
• People with no kids spend 10x as much as someone with 1 kid.

The Data

The information is a slightly modified data set taken from a Brazilian version of DoorDash® called iFood® from 2018 and is slightly modified for educational purposes (remember the data bootcamp). The source of the data can be found here and the dataset contains approximately two years of data. Some of the data set is shown below.

The rows indicate individual customers (2205) and the columns were different categories: Customer profiles, Product preferences, Campaign successes/failures, Channel performance, Size of family, etc. For a full data dictionary, please see the following image.

Business Questions

To better assess the growth of DoorDash® company and to find patterns of customers behaviors, I wanted to focus on getting answers of these business questions:

• What is the average and total amount of money spent by each customer?
• Does the income of customer affect the money spent?
• What is the age range of customers using DoorDash®?
• Does the amount of children in the home affect DoorDash® usage?

Analysis

With initial investigation into the data set, I was able to produce the following summary notes.

Thus showing us some standard aggregations to the data that we can easily make sense of. There are 2205 total customers, the oldest is 80, the newest customer has been around for 2159 days with DoorDash®. The average and sum spent is shown with this sample.

With further investigation and aggregation, a scatter plot was used to show the correlation between income and total spent. Now because you can't negatively spend, the scatter plot and the R-squared linear regression is shown above the 0 y-intercept.

From this plot, we can clearly know that 75% of the variance of spending is explained by the income. Thus, the higher the income the more you spend. Obviously there are a few outliers in this scatter. One being someone of very low income but high spending. This could possibly be explained by an anomaly, a mistake in the data, or even a student who spends a lot of their parents money on food delivery. The other outlier is "Frugal Fred" who has a very high income and does not spend a lot on food delivery. Maybe he pays for his own chef?

Following that, this histogram illustrates the amount of money people are spending on food delivery, with the average being $562 (see above). The majority of which are paying under $418.

The Bar graph shows the number of new customers that were obtained throughout the year. The overall average of new customers acquired per month is around 183 with January being the month with highest number of new customers acquired. Whereas the months of November and December, number of customers acquired were the lowest. One thought is that with the Holiday season people are saving their money during those months and giving out gift cards that are used at the beginning of the year.

Lastly, in aggregating the data deeper in Excel, I created a Pivot Table that broke down the customers by age and family size. It showed the patterns between their spending amount, their age and if they have kids or not. We can see that customers without children spent lot more money on food deliveries as compared to those who have kids. We can interpret that people with 0 kids tend to spend more money on food deliveries and were double the count of customers with 1 kid. There were only 2 customers with 2 kids. This suggests that there would need to be more data to give us a clearer understanding of the relationship between 2 kids and using food deliveries. But for now, we can estimate that it would be significantly lower than 0 or even 1 kid.

Conclusion

To recap on what's been analyzed through this dataset is that people food. In a 2018 sample, there were total of $1.24 Million spent from the food delivery company DoorDash®. We can see that 75% of the spend variance can be explained by income level. i.e the more you earn, the more you spend on food delivery. We saw DoorDash® has a steady increase of new customers throughout the year (approx. 183/mo) with the lows at the end of the year and highest at the beginning of the year. And lastly, people with 0 kids spend up to 10x more than those without kids on food deliveries.

Thank you for reading all of this. If you have any questions feel free to comment below or connect with me Brock Johnson here on LinkedIn.

I am looking for new opportunities in the data world, so if you hear of any or are in the market please reach out, thanks!