Module Doc - Twitter Volumes Evolution


This module displays the tweets volumes evolution for the given coins. This includes tweets published by the official project account, but also tweets in which the crypto was mentioned.

This may seem conter-intuitive at first, but the metric presented is not an absolute count value but is a flow value instead. It actually makes the graph simpler to read : when the curve rises it means the coin is being mentioned more often, and vice versa. The unit used corresponds to the number of tweets that would be counted in a day for this flow value, but it should only be used for relative analysis as it is based on representative samples of data.

The chart is updated every hour.

Graph Options

As all our timeline charts, this module present some handy options:

  • A tag list of your dashboard cryptos, that allow you to show only some them, if you find it too difficult to read when there's too many curves.
  • A % toggle, that switch the display of the curves from absolute values to relative values, and vice versa. The relative values for each coin are calculated compared to the first point of the curve.
  • A date range selector. Note that the date ranges available depend on the granularity of the data, as explained below.
  • A data granularity toggle. You can choose between daily granularity, which is what we use as a standard for most of our data, and what we keep for long term storage. The daily data is averaged from our raw data for the day, which granularity can vary depending on the collect process (for Twitter data collection for example we get more frequent data points for more mentioned coins). This raw data is accessible for shorter date ranges in in the "live granularity" mode.

Tweets Collection

For the tweets collection, we use the official Twitter Search API. As explained above, we check Twitter more frequently for coins that are mentioned more often, which explains the difference in number of 'live' points you can see between different cryptos.

When checking for a given crypto tweets, our number one priority is to avoid counting tweets that aren't actually mentioning the crypto. That's why we only use hashtags of the crypto name or symbol to detect tweets.

As some altcoins symbols can also be actual acronyms of other brands or products, sometimes even more popular that the altcoin itself, it is really important that we don't stop here : we've built a classifier to detect if a tweet is really about blockchain or cryptocurrencies. Each tweet relevance is thus checked before including it in our processing flow.

IMPORTANT NOTE : On 2019 August the 27th, we improved our classifier engine, in particular to be more severe about the tweet relevance. It didn't change the volumes for most of the coins, but it did reduced it drastically for a few altcoins that were benefiting from volumes of other topics sharing the same hashtags. Keep it in mind when analysing historical data.