Oranges vs Apples - No Fruit Left Unturned - Applying KPIs to Diet Part 3

When I search for “comparing apples to oranges”, I come across posts from the Smithsonian giving their nutritional facts, satire posts in scientific style, or medical journals stating they are too similar. Of course, the rest of the search results were about the phrase being an idiom to illustrate that you cannot compare things that are so different from each other. The problem with these comparisons is that it does not help in any way. As you might’ve guessed from the last two articles (part 1, part 2), we’ll discuss how you can compare them.

To compare them, regardless of how similar or different they are from each other, we need:

1. What is the goal, the job to be done, or the need?

2. What makes it easier or harder to reach the goal or meet the need?

3. How do these fruits make it easier or harder?

What is the need

1. As we discussed in part 1 of this article, fruits are a healthy snack option, which means they are probably low in GI while being low in calories, to make themselves a medium-to-high satiety snack option. So, the need is a snack option with high satiety. Although it is recommended to use any fruit to reach the 100-400 kcal per day of fruits, it bears investigation on how the metrics across fruits vary.

2. What will make it easier or harder to gain satiety? The slower the digestion of the fruit, the higher the satiety.

3. What part of fruit helps? The more the fiber, the more the water content, the less the fast-digesting carbohydrates vs polysaccharides, the better the satiety.

Now that we uncovered the 3 questions we started with, let’s look at the metrics.

Understanding the Fruits

Learning from our foray into vegetables in part 2 of our article, some potential metrics and counter-metrics valuable for fruits are:

1. Amount of Fiber (g in 100g of fruit)

2. Amount of Calories (kcal in 100g of fruit)

3. Amount of Sugar (g in 100g of fruit). I’ve used sugar instead of carbohydrates for fruits since most fruits did not have long-chain carbohydrates.

4. “Goodness” i.e. Fiber/Calorie (g/kcal). We know this factor might be handy from our investigations into vegetables.

We will not use GI (wiki for Glycemic index) to measure the fruits because GI can be misinterpreted for cases such as watermelon where GI is high suggests not a good fruit whereas watermelon has a lot of water content so in 100g of watermelon the sugar gets diluted sufficiently. Due to such examples, my readings suggested that GI is not a metric that is individually sufficient to rank foods.

The calories per gram in fruits vary significantly, ranging from -44% to +52% from the average for sugar-based fruits or +174% for avocado. What if we compare the amount of sugar in fruit with the number of calories in the fruit? Please ignore the “SUM of” artefacts of using pivot tables.

It’s interesting that even excluding Avocado, the sugar amount and calories amount do not correlate. This might be because fiber is also another type of carbohydrate that contributes to the total calories. On combining sugar and fiber to compare with the total calories, there is a lot more correlation.

This suggests that for the same amount of calories, getting more fiber will likely mean less sugar. So, we come back to the same “goodness” metric we used for vegetables - fiber/calorie ratio. The fiber/calorie metric ranges from +145% to -73% from the average which is a large range. Another benefit of this metric is that avocado is no longer an outlier.

This metric can suggest the “goodness” across fruits. Given this, I nowadays choose to buy fruits that are high in this metric and have reduced my purchases of fruits low on this metric. So, we had more success in ranking fruits than vegetables, because across fruits we found a single metric that could be used to rank them for a hypocaloric diet.

Conclusion for Fruits

Given a goal (hypocaloric diet), we can identify steps to help it (high satiety diet), and find metrics that help step forward (fiber/calorie). Through a few trials with data and connecting the dots between the data and what we infer to be right, we could shortlist the right metric to rank the fruits.

And here is the ranking:

Comparing Apples to Oranges

Across all the metrics that oranges are ranked higher than apples. So, Oranges are better than apples!

Granted, not by much.

Comparing Fruits to Vegetables

After having discussed in part 1 the reasons that fruits or vegetables were in the diet plan and having reviewed metrics for it, here are some hypothesis to test for fruits vs vegetables:

1. Fruits are much more calorie-dense than the recommended vegetables

2. Vegetables have more fiber than fruits in each 100g

3. Vegetables are higher than fruits in fiber/calorie because, for every 100g, if they have higher fiber and lower calorie, their fiber/calorie is much higher than fruits.

Let’s test this out.

The broader trend we see is fruits have a higher calorie-density than recommended vegetables. It is also surprising that the exception vegetables are as calorie-dense as some of the sweetest fruits. This makes the comparison more vivid to think that potatoes are more calorie-dense than sweet sweet mangoes or juicy grapes. Sweetness and calorie-density aren’t well-connected, but it was interesting to note this.

Fiber distribution has much less pattern to it than calorie distribution. What about fiber/calorie distribution? We can look at the distribution in two ways, one is using the normalized metric, and another is using spread across two-dimensions.

The spatial distribution is an interesting way to understand the distribution. One interesting insight the 2D distribution provides that wasn’t understood in the 1D distribution is that there are some vegetables where both fiber and calories are close to 0. I’m not sure what is the benefit of eating those beyond micronutrients. Looking at the distribution, a change I’ve considered in my daily diet is to stop putting in the money and effort to buy and chop iceberg lettuce for a salad with fruits - it doesn’t seem to be of value.

Takeaways

I hope the use of data to define metrics towards a goal has helped you think of new ways of measuring impact (more in part 1). Do you have a new perspective for using KPIs beyond business? For using visualizations to understand data? Do you feel like you have a cheat sheet of fruit rankings now to compare apples, oranges, and so much more?

What other things out there might be interesting to measure?