Regarding Tweet_Volumes Results in Rest API

[vigonjeltz]

I recently used twitters trends using the following for running some basic analytics

https://dev.twitter.com/rest/reference/get/trends/place

which Returned the top 50 trending topics for a specific WOEID. I used the “world” and “USA” just for testing and noticed that for the top 50 trends in the response the following occurred.
* for USA only 14 tweet_volumes had data and the rest 36 were Null
* for WORLD only 31 tweet_volumes had data and the rest 19 were Null

Why are tweet volumes not present for all the trends ? Is this common ?

[andypiper]

Tweet volumes are a new addition to the API responses (added about a month ago), and as a result I can’t comment on exactly what the expected behaviour is / how common this would be without checking in more detail. I’ll try to do that.

[vigonjeltz]

Andy,
Thank you so much for your response. I was unaware that “tweet-volumes” was added recently . thanks again and keep me posted.

Regards

vick

[CHCJoski]

Are there any plans to improve this functionality? I’m working on a capstone project for school and I need to analyze the most popular trends, and quantify that data. I need to know just how much more popular one trend is than another, and without the “tweet_volume,” I’m at a loss on how to do that.

Does anyone know another way I can use the API to get the most popular trends and consistently get data that shows the trends? Or a way to get that data not using the API? Does Twitter itself have a webpage that lists all of the trends and their tweet volume?

[Donkey]

It seems that below a value of 10,000, tweet_volumes is always null. However, on Twitter website, real volumes are indicated.

[t0_wire]

Thank you for adding this information. I have also just run across this issue, after finding that only six of the fifty trends returned contained tweet_volume data.

The API (https://dev.twitter.com/rest/reference/get/trends/place) very clearly states:

The tweet volume (tweet_volume) for the last 24 hours is also returned for every trend.

It’s difficult to understand why the implementation varies from Twitter’s own API documentation, and why the threshold is so arbitrary. The data must be available for all of them, otherwise they wouldn’t have been identified as trends in the first place… so it’s difficult to see how there could be a significant overhead for including the data. It’s a shame there was no further update from Andy.

I really hope Twitter looks at this again, as it prevents developers from doing any really interesting work of their own on trends that might add to what is available from Twitter’s own interface.

[Donkey]

Are you sure about that:

only six of the fifty trends returned contained tweet_volume data

?
I have a sample where 13 tweet_volume contain a number (trends for France (23424819)).

I think Twitter keeps the privilege to have more information than what it can offer to developers…

[andypiper]

With numbers of this kind, there tends to be a threshold below which the platform does not report specific figures. Also, note that the numbers of trends with volume data may vary based on the location chosen for trends reporting from the API.

[t0_wire]

It would be helpful if this information was added to the API documentation, so that developers do not waste their time writing code to use data that isn’t available.

[andypiper]

Thanks for the feedback - I’ll look at getting that updated.

[lyle_city]

Hello,

Any update regarding this topic ?
Thanks a lot in advance !

Adrien

[andypiper]

What specific update are you looking for beyond what I provided in my most recent response? Thank you.

[lyle_city]

Hi Andy ! Looking at your last answer September, 14th 2016, you were looking at getting this tweet_volume issue updated.

Concretely, calling the API on the trends/place, we still get a lot (a vast majority) of results with tweet_volume: null. Did I miss a good practice or is it to be corrected ?

Kind Regards

[andypiper]

That’s to do with thresholds as I mentioned in a previous answer. We should get the documentation updated to specify that, but I’m not aware of any plan to do anything further.

[lyle_city]

Thanks a lot for your quick and relevant answers, Andy. We’re a bit disappointed to see that the tweet_volume is not 100% accurate but will do our best to find a solution to identify main activities in a specific WOEID. Have a good day !

[hashgear_shop]

You mentioned the threshold below which the API returns null – but is there a threshold for inclusion in the trends list in the first place? For instance, do we know that there are at least 5000 tweets?

In other words, what is the threshold for trending for any given location? Is it calculable given the data from other trending topics?

Thanks,
Marc

[andypiper]

We’re unable to describe the specifics of internal algorithms in these cases, apologies. There are no current plans to further update the trends endpoints.