{
  "fetched_at": "2026-10-20T01:57:37Z",
  "source_url": "https://www.kaggle.com/competitions/airbus-finance-ml-workshop/data",
  "pages": [
    {
      "name": "data-description",
      "content": "The Twitter airline dataset is a collection of comments posted by Twitter users about their travel experiences with different airlines. The dataset is composed of a total of 5,313 tweets that were gathered over a period of six months.\n\nEach tweet has been labelled as either positive or negative based on the sentiment of the tweet towards the airline. \n\n## The data has been split into training (3,963 rows) and testing (1,350 rows) datasets\n\n*   **airline_tweets_train.csv** - The training set, we will use this dataset to train our models\n*   **airline_tweets_test.csv** - The test set, we will evaluate and rank teams with this dataset\n\n## Columns\n\n*   `Id` - Numeric Id of the tweet\n*   `airline_sentiment` - Sentiment of the tweet, 'positive' 'negative' or 'neutral'\n*   `airline` - Name of the airline referenced in the tweet\n*   `text` - Content of the tweet\n*   `retweet_count` - Number of retweets performed on tweet\n*   `user_timezone` - Timezone of the user posting tweet\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "## Competition\n\nWelcome to the **Airbus Finance NLP Kaggle Competition** on classifying Twitter comments regarding airlines. \n\n* Participants will be provided with a dataset of tweets related to various airlines, along with their sentiment labels (positive or negative)(airline_sentiment_train.csv)\n* We will process the text using common **Natural Language Processing (NLP)** techniques\n* We will train a **ML Classification Model**\n* Participants will also be provided with a dataset of tweets without a sentiment label (airline_sentiment_test.csv)\n* Each team will submit their predictions and see their ranking in the **Leaderboard** section\n\n\n## Python Code:\n* A working code will be provided, only minor adjustments required\n* The aim of this session is not to completely understand the code, but what the code is performing\n* Don't worry if some parts of the code are unclear\n\nGood luck, and we look forward to seeing your sentiment analysis!\n\n## Prize\nTo format pages, stick to the following conventions:\n- Inline `code` should be encapsulated by back-ticks \\``. \n- Block code should have  \\``` as the first and last line of the block to produce\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "The evaluation metric used will be Accuracy: The number of instances correctly classified divided by the total number of instances.\n\nIn this case our dataset is very balanced (same number of positives and negatives) so this metric is an appropriate one.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "Rules:\n - Have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}