Generating random noise for media data
\n \t
\n
<\/BR>
\n
\n \t
\n
<\/BR><\/p>\n
Hey everyone – I work on an ML team in the industry, and I\u2019m currently building a predictive model to catch signals in live media data to sense when potential viral moments or crises are happening for brands. We have live media trackers at my company that capture all articles, including their sentiment (positive, negative, neutral). <\/p>\n
I currently am using ARIMA to predict out a certain amount of time steps, then using an LSTM to determine whether the volume of articles is anomalous given historical data trends. <\/p>\n
However, the nature of media is there\u2019s so much randomness, so just taking the ARIMA projection is not enough. Because of that, I\u2019m using Monte Carlo simulation to run an LSTM on a bunch of different forecasts that incorporate an added noise signal for each simulation. Then, that forces a probability of how likely it is that a crisis\/viral moment will happen.<\/p>\n
I\u2019ve been experimenting with a bunch of methods on how to generate a random noise signal, and while I\u2019m close to getting something, I still feel like I\u2019m missing a method that\u2019s concrete and backed by research\/methodology. <\/p>\n
Does anyone know of approaches on how to effectively generate random noise signals for PR data? Or know of any articles on this topic?<\/p>\n
Thank you!<\/p>\n<\/p><\/div>\n
submitted by \/u\/Entire_Island8561 <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span>\n<\/div>\n