# Applications of Deep Learning (WUSTL,Fall 2022) — Data

来源：https://www.kaggle.com/competitions/applications-of-deep-learning-wustlfall-2022/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

## Files

*   **sample_submit.csv** - Sample submission file.
*   **sales_test.csv** - The test/submit data for the competition.
*   **sales_train.csv** - The training data (tabular) for the competition. Sales over time.
*   **items.csv** - The 100 items for sale, includes textual NLP data.
*   **resturants.csv** - The restaurants that sell the items.
*   **./cam/*.jpg** - Daily images from a webcam placed near the restaurants.


## Columns

Items: 

*   `id` - Unique identifier for this item. The item ID.
*   `store_id` - The restaurant that sells this item. Restaurant ID.
*   `name` - The name of this item. NLP techniques are likely helpful with this item.
*   `kcal` - How many calories is this item.
*   `cost` - The base cost for this item.

Restaurants:

*   `id` - Unique identifier for this restaurant. The Restaurant ID.
*   `name` - The name of the restaurant.

Sales (training)
*   `date` - The date that this sale occured.
*   `item_id` - The item that was sold. The item ID.
*   `price` - The price that this item was sold for on this day.
*   `item_count` - Unique identifier for this item. The target (y).

Submission File:
*   `id` - Unique identifier for this prediction, aligns to the same `id` field from the test data.
*   `item_count` - The prediction (y-hat).

## Tabular Data

The tabular data for this competition exists in the sales, items and restaurants CSV files.

## Image Data

There is a webcam that records an image a day in front of the restaurants. You can see sample images here. Notice the size of the crowd on the street and beach. The filenames correspond to the date the image was taken.

![Beach CAM 1](https://data.heatonresearch.com/images/wustl/kaggle/demand_forecast/2019_01_01.jpg)

![Beach CAM 2](https://data.heatonresearch.com/images/wustl/kaggle/demand_forecast/2019_01_02.jpg)

![Beach CAM 3](https://data.heatonresearch.com/images/wustl/kaggle/demand_forecast/2019_01_06.jpg)

## Textual Data

There is also textual data in the items files. This data might also be useful.