𝗢𝗻𝗲 𝗖𝗹𝗮𝘀𝘀 𝗖𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗦𝘂𝗿𝘃𝗲𝘆

𝗢𝗻𝗲 𝗖𝗹𝗮𝘀𝘀 𝗖𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗦𝘂𝗿𝘃𝗲𝘆

You have data for one group. You need to find outliers. This is One Class Classification. It finds fraud or system errors.

You train your model on normal data. The model learns a pattern. New data arrives. The model checks the fit. Non-fitting data is an anomaly.

Common methods:

• One Class SVM
• Isolation Forest
• Autoencoders

Source: https://dev.to/paperium/one-class-classification-a-survey-36mh Optional learning community: https://t.me/GyaanSetuAi