The dataset contains telelogs - automatically generated fault and event logs from network equipment. Each record in the training set includes the corresponding root cause, while the test set contains telelogs only. Your task is to identify the most likely root cause for each test record.

A possible approach to construct further synthetic data and fine-tune off-the shelf models is described here.

This challenge provides real-world 5G telecommunication drive-test logs paired with engineering parameters from nearby network cells.

Each training example consists of:

1. A question describing a real network performance degradation scenario
2. A drive-test data excerpt (user-plane measurements across time and location)
3. A set of engineering parameters for relevant 5G cells
4. The ground-truth root cause label among eight possible fault categories

Your task is to build a model that can understand the interaction between radio KPIs, cell configurations, mobility behaviour, and coverage/topology - and correctly predict the most likely root cause of low throughput events.