Hello, assuming 1 FGS and 1 AIRS spectrum per planet, my current code takes ~5 mins per planet to produce a single row in the submission.csv. My submission times out so assuming 9h runtime --> 108 planets in the hidden test set. From those who have successfully submitted, what's your estimate on the number of planets (or spectrums) in the hidden test set? Or equivalently, what's the upper bound on processing time (assuming Kaggle's notebook resources) per planet to target when designing a solution?
", "messages": [ { "id": 3282572, "postDate": "2025-09-06T13:44:10.790Z", "content": "Hello, assuming 1 FGS and 1 AIRS spectrum per planet, my current code takes ~5 mins per planet to produce a single row in the submission.csv. My submission times out so assuming 9h runtime --> 108 planets in the hidden test set. From those who have successfully submitted, what's your estimate on the number of planets (or spectrums) in the hidden test set? Or equivalently, what's the upper bound on processing time (assuming Kaggle's notebook resources) per planet to target when designing a solution?
", "rawMarkdown": "Hello, assuming 1 FGS and 1 AIRS spectrum per planet, my current code takes ~5 mins per planet to produce a single row in the submission.csv. My submission times out so assuming 9h runtime --> 108 planets in the hidden test set. From those who have successfully submitted, what's your estimate on the number of planets (or spectrums) in the hidden test set? Or equivalently, what's the upper bound on processing time (assuming Kaggle's notebook resources) per planet to target when designing a solution?", "votes": 1 }, { "id": 3282575, "postDate": "2025-09-06T13:50:09.627Z", "content": "https://www.kaggle.com/competitions/ariel-data-challenge-2025/data
\n\n\nExpect to see roughly 1,100 exoplanets in the hidden test set.
\n
The number fits well with expected runtime and I wouldn't be suprised if it is actually exactly 1100 planets in test. Note that some of these planets have two observations. So if you want to make use of that, you will have to allocate extra runtime for that.
", "rawMarkdown": "https://www.kaggle.com/competitions/ariel-data-challenge-2025/data\n\n> Expect to see roughly 1,100 exoplanets in the hidden test set.\n\nThe number fits well with expected runtime and I wouldn't be suprised if it is actually exactly 1100 planets in test. Note that some of these planets have two observations. So if you want to make use of that, you will have to allocate extra runtime for that.", "votes": 2, "replies": [ { "id": 3282826, "postDate": "2025-09-06T21:05:37.113Z", "content": "Ah thank you! I missed that 🤦
", "rawMarkdown": "Ah thank you! I missed that 🤦" } ] } ], "comments": [ { "id": 3282575, "author_name": "Pascal Pfeiffer", "author_url": "", "post_date": "2025-09-06T13:50:09.627000", "content": "https://www.kaggle.com/competitions/ariel-data-challenge-2025/data
\n\n\nExpect to see roughly 1,100 exoplanets in the hidden test set.
\n
The number fits well with expected runtime and I wouldn't be suprised if it is actually exactly 1100 planets in test. Note that some of these planets have two observations. So if you want to make use of that, you will have to allocate extra runtime for that.
", "votes": 2, "replies": [ { "id": 3282826, "author_name": "Nishan Mann", "author_url": "", "post_date": "2025-09-06T21:05:37.113000", "content": "Ah thank you! I missed that 🤦
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "3282572": "Hello, assuming 1 FGS and 1 AIRS spectrum per planet, my current code takes ~5 mins per planet to produce a single row in the submission.csv. My submission times out so assuming 9h runtime --> 108 planets in the hidden test set. From those who have successfully submitted, what's your estimate on the number of planets (or spectrums) in the hidden test set? Or equivalently, what's the upper bound on processing time (assuming Kaggle's notebook resources) per planet to target when designing a solution?", "3282575": "https://www.kaggle.com/competitions/ariel-data-challenge-2025/data\n\n> Expect to see roughly 1,100 exoplanets in the hidden test set.\n\nThe number fits well with expected runtime and I wouldn't be suprised if it is actually exactly 1100 planets in test. Note that some of these planets have two observations. So if you want to make use of that, you will have to allocate extra runtime for that." } }