## Metric
The evaluation metric for this competition is *accuracy*. The metric is calculated as:

$$accuracy = \frac{TP+ TN}{TP+TN+FP+FN}$$

Where:

- TP is *true positive*, positive cases predicted as positive
- TN is *true negative*, negative cases predicted as negative
- FP is *false positive*, negative cases predicted as positive
- FN is *false negative*, positive cases predicted as negative
- TP+TN+FP+FN is equivalent to the number of cases predicted

As we have four targets here (`q_noise_nearby`, `q_noise_kind`, `q_earphones`, `q_thermal_preference`), accuracy for each target column is calculated and the average of the four of them is returned.

## Notebook Submissions
You can make submissions directly from Kaggle Notebooks. By adding your teammates as collaborators on a notebook, you can share and edit code privately with them.

## Submission File
For each `id_unique` in the test set, you must predict the target variables: `q_noise_nearby`, `q_noise_kind`, `q_earphones`, `q_thermal_preference`. The file should contain a header and have the following format:

```python
id_unique,q_noise_nearby,q_noise_kind,q_earphones,q_thermal_preference
1,None,Not applicable,Not applicable,Cooler
2,A little,Appliances,No earphones,No change,Public
3,A little,Talking,No earphones,Cooler
```