Alert System for atypical variations

AI Discussions
,
1

I’m working on an alert system for atypical variations in a specific product. The idea is to train an AI model with historical data of this product and then monitor it continuously, alerting me to abnormal fluctuations. Does anyone have suggestions on how to tackle this issue or know of articles that could provide more information for development?

I’m in the early stages, setting up the foundation. However, I’m unsure about which algorithms would perform well for this task.

Thank you in advance!

2

u can try searching for “abnomally detection algorithm”. The key idea is using normal distribution

1 Like
3

Indeed, anomaly detection.

1 Like
4

Hello, first of all, I would like to thank you for your response. I am currently studying the algorithm and considering how to apply it to my case. However, I’m curious to know if there is anything that takes time into consideration. I would like to establish a price window - with maximum and minimum values - determined from the historical data. With this approach, any value outside of this window would be considered an anomaly.

Additionally, I’ve been contemplating the idea of assigning weights to features - as price is not the sole relevant factor in my model - and generating an alert score. However, I’m interested to know if there are any AI techniques that could contribute to this approach. Does anyone have any suggestions in this regard?

Thank you!

5

There is a body of work on this topic.

For example, a quick internet search for “anomalies in time series data” turned up this one:

1 Like