We have improved the way our ‘Top’ rankings for categories are calculated. We now represent more accurately the most and least loved topics within the category.
Whereas before, the topics with fewer positive opinions could still maintain a high ranking, we have now changed the formula to also take into account the volume of messages received.
In short, our method calculates the average love percentage across the whole category, against the volume of messages and percentage love for a specific topic.
Check out some examples of our fancy new rankings here:
Want to know more? Read on…
We are using a well-known method called Bayesian average to calculate a love score that we use to rank topics within a category.
The idea is to estimate the love score of each topic based not only on the proportion of positive and negative opinions; but also including the volume of those opinions (within the period).
This method accounts for irregularities like topics with few comments (and mostly positive) or topics with large amounts of opinions.
For example, in a calculation of an average love score of a topic where only two positive opinions are available, a normal average score would be 100%. However, as only two opinions are available, 100% may not represent the true average had more reviews been available. We instead calculate a Bayesian average of this score by adding the average score of all topics in the category to the calculation.
If you want to know more details, please don’t hesitate to contact us.