What does Amazon do? Part II

Loading...

From the course by Princeton University

Networks Illustrated: Principles without Calculus

39 ratings

Princeton University

Networks Illustrated: Principles without Calculus

39 ratings

What makes WiFi faster at home than at a coffee shop? How does Google order its search results from the trillions of webpages on the Internet? Why does Verizon charge $15 for every GB of data we use? Is it really true that we are connected in six social steps or less? These are just a few of the many intriguing questions we can ask about the social and technical networks that form integral parts of our daily lives. This course is about exploring the answers, using a language that anyone can understand. We will focus on fundamental principles like “sharing is hard”, “crowds are wise”, and “network of networks” that have guided the design and sustainability of today’s networks, and summarize the theories behind everything from the social connections we make on platforms like Facebook to the technology upon which these websites run. Unlike other networking courses, the mathematics included here are no more complicated than adding and multiplying numbers. While mathematical details are necessary to fully specify the algorithms and systems we investigate, they are not required to understand the main ideas. We use illustrations, analogies, and anecdotes about networks as pedagogical tools in lieu of detailed equations. Please note that per Princeton University policy, no certificates, credentials or reports are awarded in connection with this course.

From the lesson

Product Rating on Amazon

The decision of whether or not to purchase something online is often driven by the ratings that previous customers have left for it. In this lesson, we will take a look at Amazon's review system, and the see how "crowds are wise" is another important networking principle.

Amazon & eCommerce4:28

Average Ratings10:21

The Wisdom of Crowds10:03

Rating Aggregation Challenges4:11

Naive Averaging5:57

Bayesian Ranking: Part I5:32

Bayesian Ranking: Part II9:14

Bayesian Ranking in Practice5:41

What does Amazon do? Part I5:00

What does Amazon do? Part II5:29

Meet the Instructors

Christopher Brinton
Lecturer
Electrical Engineering
Mung Chiang
Professor
Electrical Engineering

First let's apply Bayesian adjustment to this and see what happens.

In order to use Bayesian adjustment as we discussed, we

need to be able to compute first the overall average.

Doing that is pretty straightforward in this case.

We're just going to take the average rating overall all the products.

Here it's assumed to be the top 20, right?

And, that's not necessarily the valid assumption

to make, because it, you know, including the

top 30, top 40 could change it up, but we'll assume it's the top 20 here.

And that gives an average rating of 4.59 When

we add up the total number of stars and

find the total number of reviews submitted and do

the average rate and computation, for the overall average.

The second thing, which is a little more

difficult, to, to choose what overall num is.

Alright, because as we saw before, we don't necessarily just want

to use, the total number of reviews that were submitted for that.

We may want to use, something instead, like, what Beer Advocate does, which

is, to use the minimum number of reviews required to submit on the site.

So we will try out a few different cases.

One is the lowest number of reviews entered

for any product in the list, which is 19.

Second is the highest number of reviews for

any product in the list, which is 1,064.

Third is the average number of reviews per product right, so if you

just take the average, total I have a number of reviews entered, that's 235.

And fourth is the total number of reviews which

would just be using the standard definition of overall num.

Which is 4,704.

So, if we do that out in each case,

usually the four cases, and, we rank the lists accordingly.

The first case we'll get, 17 followed by 1, 2, 15 and so forth.

So, this is the rank order list.

The second case will have 1, 2, 1, 3, 4, 2, 17, 15, and so forth.

Third case something very similar to that.

And in the fourth case something a little different again.

So if you look at these quantitatively and compare them they,

the first and the last ones seem to be pretty off.

I mean they, you know, a lot of these

are pretty random in terms of what our objective

is and, to remind you our objective is to

get to the rank-order list that's put on Amazon.

And that would be 1,2,3,4,5,6, blah, blah, blah down to 20.

However the second two right here, the highest

number of reviews for any product, 1064, and

the average number of reviews, 235, So you

could do a little better, you know 1,3,4,2.

In terms of at least of having, similar clusters of products grouped together.

And as we said, the highest and the average seem to give

better results, even though there are still clear exceptions in each case.

And, to try to understand this we'll focus on products 8, 15, 17 and 20.

Try to see where the ranks where they are.

But for products 15, 17 and 20, the question is why are they put so low on the

list and, additionally, why are they scattered from one

another given that they're put so low on the list.

Well they each of these have really the lowest number of ratings.

So these, these are the products that had

the lowest number of ratings entered for them.

So that could explain why they're placed lower and demoted lower onto the list.

And with Amazon, this case really demotes products for being lower on the list.

But that still doesn't explain why they're

scattered from one another because if they have

the lowest number of ratings then why aren't they just placed at 18, 19 and 20?

So if we look at product 15 however one thing we find is

that 15's most helpful review had 60 of 61 people find it helpful.

Which is good.

And the most helpful review is also very positive.

So, they've got five star and 60 out of 61 found that review helpful.

On the other hand product 17 and 20 only had seven and six people find

their top reviews helpful respectively, which could explain

why 15 is ranked higher in this list.

So this could explain why 15 is pushed up in the ranking.

Now the question is, why are 17 and 20 placed far apart?

Well, if we look at product 17, product 17's

most helpful review was more recent, occurring in 2012.

Whereas, product 20's most helpful review was less, recent.

So this should be, more recent.

And so that could be at saving grace is the

fact that, it had a very helpful review occurring, recently.

And notice also that despite being ranked so low, product

17 is actually the one that has the highest average rating.

So in addition to what we just discussed, there

are two more factors that could help explain this.

First is that only half the reviews are from 2012, and second is that the average

review reputation is fairly low, with all but

two having less than ten helpful votes overall.

And finally we should point out that product 15

had an extremely high quality review and its descriptive

listings in the pros and cons of each case

and this also could have pushed its rank higher.

The next question we want to understand is why was product eight placed

so high, given that its Bayesian ranking puts it around the 16th position?

Well it turns that item 8's most helpful

review had 79 of 79 people find it helpful.

Additionally, over 10 of the top 5000

reviewers on Amazon had actually reviewed this product.

And, rankings on Amazon extend to around 3 million people over 3 million people.

Finally, one of the reviewers was a Hall of

Fame Reviewer and he had given it five stars.

So, when you're on Amazon, you can actually tell if the person

a Hall of Fame Reviewer or not, meaning they're in top ten.

And that person had given this five stars.

So each of these may have caused product 8 to be severely promoted

in the rankings because quality of reviews can override the quantity of reviews.

[BLANK_AUDIO]

Explore our Catalog

Join for free and get personalized recommendations, updates and offers.

Coursera provides universal access to the world’s best education, partnering with top universities and organizations to offer courses online.

© 2018 Coursera Inc. All rights reserved.

Download on the App Store

Get it on Google Play