SQL Server + C# - Naive Bayes modelling proof of concept

We are researching a project to determine and model user behaviour, and are looking for a proof of concept implementation. The goal is to watch login activity for users, build up a model of "what's normal", and be able to assign a score to each login (how far out of normal score).

For example, if I normally login from ip 1.1.1.1, on weekdays, and in the morning - and then I login from a different IP - that login should receive a lower score.

This should leverage the Naive Bayes modelling support inside of sql server.

<[url removed, login to view]>

We will provide login sample data, and the proof of concept should build a model based on the following factors:

- source ip address

- time of day (4 hour buckets, midnight-4am, 4am-8am, etc)

- day of week (mon, tue, etc)

工程 软件构架 软件测试

项目ID： #2705022

关于项目