If you're not familiar, Frank's 2000" TV is the title of a song from "Weird Al" Yankovic's 1993 album Alapalooza. In which the song's titular Frank buys (you guessed it) a 2000" TV.

So I decided I'd try my hand and figuring this out based on data I could glean from current TV sizes and prices. To do this I took used the Best Buy API to pull down all of the televisions they have in their database and then get the price, weight and size¹. Then taking that data I transformed it into two data sets, price vs size and weight vs size. Then taking that data I inserted it into Highcharts using a regression library² to get a best fit line. I tried a couple of different line models until I found what I think is the best fit³.

To get the data from the Best Buy API I made two queries (five if you count the ones used to get the pagination results)

List all categories

curl "http://api.remix.bestbuy.com/v1/categories?format=json&apiKey=$API_KEY&show=id,name"

List all products in that category

curl "http://api.remix.bestbuy.com/v1/products((search=tv)&categoryPath.id=abcat0101000)?show=name,sku,salePrice,details.name,details.value&format=json&pageSize=100&page=1&apiKey=$APIKEY"

Best fit lines

After pushing the data into Highcharts and running the regression library² I got the following formulas

• Price — y = 1.59x2 + -85.52x + 1407.76
• Weight — y = 0.03x2 + -0.91x + 15.26

¹ Since I did this in an automated fashion some of the entries were not formatted consistantly and so I threw them out. For example some of the sizes were denoted as 54-1/2 instead of 54.5 and rather than code up all the possibilities, I just removed them

² The regression charting library was developed by Ignacio Vazquez and can be found on GitHub

³ I am not a mathmatician and I've not taken statistics in a long time. I only tried a couple of different formulas and eyeballed their matches