We provide a large training dataset consisting of year-long hourly electricity meter readings from 200 buildings. Each datapoint in the meter readings is annotated whether it is anomalous (`1`) or normal (`0`) usage.  Your task is to develop accurate anomaly detection models using this training dataset and then predict whether each each meter reading in the test dataset (from another 206 buildings) is anomalous (`1`) or normal (`0`).

## Files

**train.csv** - The training set

* `building_id` - Unique building id code. 
*  `timestamp` - When the measurement was taken
*  `meter_reading` - Electricity consumption in kWh.
*  `anomaly` - Whether this reading is anomalous (`1`) or not (`0`).

**test.csv** - The test set

* `row_id` - Unique row id for your submission file.
* `building_id` - Unique building id code. 
*  `timestamp` - When the measurement was taken
*  `meter_reading` - Electricity consumption in kWh.
 
**sample_submission.csv** - Sample submission file

* `row_id` - Unique row id.
* `anomaly` - `1`(anomaly) or `0`(normal)

**train_features.csv** and **test_features.csv** - List of [features](https://github.com/buds-lab/ashrae-great-energy-predictor-3-solution-analysis/blob/master/solutions/rank-1/scripts/02_preprocess_data.py) used by the [winners](https://www.kaggle.com/competitions/ashrae-energy-prediction/discussion/124709) of the [ASHRAE - Great Energy Predictor III](https://www.kaggle.com/competitions/ashrae-energy-prediction/overview). These are additional features about the buildings and meter readings that might be useful to improve the model prediction.