If I told you I had a banana at 9am yesterday, could you predict what else I had? Would you have said coffee and oatmeal? Congratulations, you just won your first spin on… the meal wheel!
The latest version of the UP app, released today, contains an entirely new food tracking interface designed to help you log meals easily and quickly. One way to make food logging easier? Magic. Or… data science. UP will now predict and suggest the most likely meal companions for a single food item, saving you time.
This is where the Jawbone UP community comes in. Using the foods people have logged in the past, I computed the probability that two foods occur in the same meal. And of course, certain foods are more likely to appear together than others. Chicken and rice? Sure. Bacon and peanut butter? Not likely… unless you’re an adventurous eater.
Morning Meal Wheel
In addition to the most likely food items, I also take into account the time of day your meal is logged. If you’re eating spinach in the morning, it’s more likely you enjoyed it in an omelet. If you’re eating spinach at 6pm, you may have had it in a salad. Find spinach in both the morning and afternoon/evening wheels and see for yourself!
Afternoon/Evening Meal Wheel
This feature may also remind you to log items you may typically forget, like ketchup, which can affect the overall healthiness of a meal.
Play around with the meal wheels, and discover food favorites of the UP community. Who knew ice cream and french fries went together like peas and carrots? I thought it was just me!
I extract common food pairings by examining anonymized meals logged by UP wearers through June 2014. To suggest the next food item to log, I calculate the conditional probability of that item given other items already logged. Extremely rare foods and pairings are excluded. The visualization above only includes a sample of the most frequently logged foods.