![\"old](\"https://i.imgur.com/hQqp1m7.png\")
I reran all the EDA I did previously on the new data and I could not find the statistical differences I once found. I hope we're safe now.
\nHere is an example. Top picture is from original data, bottom one is from new data. x is sum of mean power for snippets 0,2,4 and y is sum of mean power for snippets 1,3,5. Power is the element wise squared image. Red is target 1.
\n
![\"new](\"https://i.imgur.com/Gz073Xc.png\")
Given mean is almost zero, power is almost the same as variance. We see that on new data variance is distributed the same way for each target values. It was not the case with original data.
\nEdit. Some reported they can't see images. They are at:
\nhttps://i.imgur.com/hQqp1m7.png
\nhttps://i.imgur.com/Gz073Xc.png
", "messages": [ { "id": 1391279, "postDate": "2021-07-17T13:09:52.350Z", "content": "I reran all the EDA I did previously on the new data and I could not find the statistical differences I once found. I hope we're safe now.
\nHere is an example. Top picture is from original data, bottom one is from new data. x is sum of mean power for snippets 0,2,4 and y is sum of mean power for snippets 1,3,5. Power is the element wise squared image. Red is target 1.
\n
Given mean is almost zero, power is almost the same as variance. We see that on new data variance is distributed the same way for each target values. It was not the case with original data.
\nEdit. Some reported they can't see images. They are at:
\nhttps://i.imgur.com/hQqp1m7.png
\nhttps://i.imgur.com/Gz073Xc.png
", "rawMarkdown": "I reran all the EDA I did previously on the new data and I could not find the statistical differences I once found. I hope we're safe now.\n\nHere is an example. Top picture is from original data, bottom one is from new data. x is sum of mean power for snippets 0,2,4 and y is sum of mean power for snippets 1,3,5. Power is the element wise squared image. Red is target 1.\n\n\n\n\n\nGiven mean is almost zero, power is almost the same as variance. We see that on new data variance is distributed the same way for each target values. It was not the case with original data.\n\nEdit. Some reported they can't see images. They are at:\n\nhttps://i.imgur.com/hQqp1m7.png\n\nhttps://i.imgur.com/Gz073Xc.png\n\n", "votes": 32 }, { "id": 1392448, "postDate": "2021-07-18T17:39:46.417Z", "content": "Agreed! I ran a Kolmogorov-Smirnov / Malmgren-Svensson / Cramér-von Mises, and they are now all within error bars. Before, it would look like the \"wild west\"
", "rawMarkdown": "Agreed! I ran a Kolmogorov-Smirnov / Malmgren-Svensson / Cramér-von Mises, and they are now all within error bars. Before, it would look like the \"wild west\"", "votes": 1 }, { "id": 1391420, "postDate": "2021-07-17T15:20:22.443Z", "content": "sorry why I cannot see the figure?
\n\n", "rawMarkdown": "sorry why I cannot see the figure? \n> I reran all the EDA I did previously on the new data and I could not find the statistical differences I once found. I hope we're safe now.\n> \n> Yop picture is form original data, bottom one is from new data. x is mean of power for snippets 0,2,4 and y is mean of power for snippets 1,3,5. Power is the element wise squared image. Red is target 1.\n> \n> \n> \n> \n> \n> Given mean is almost zero, power is almost the same as variance. We see that on new data variance is distributed the same way for each target values. It was not the case with original data.\n\n", "votes": 1, "replies": [ { "id": 1391492, "postDate": "2021-07-17T16:42:24.363Z", "content": "I reran all the EDA I did previously on the new data and I could not find the statistical differences I once found. I hope we're safe now.
\nYop picture is form original data, bottom one is from new data. x is mean of power for snippets 0,2,4 and y is mean of power for snippets 1,3,5. Power is the element wise squared image. Red is target 1.
\n\n
\n
Given mean is almost zero, power is almost the same as variance. We see that on new data variance is distributed the same way for each target values. It was not the case with original data.
\n
I can see it in your comment. What browser do you use? If you open the link to the image can you see it?
", "rawMarkdown": "I can see it in your comment. What browser do you use? If you open the link to the image can you see it?\n", "votes": 1 }, { "id": 1392971, "postDate": "2021-07-19T09:12:50.727Z", "content": "I cannot see it too. I use chrome btw
", "rawMarkdown": "I cannot see it too. I use chrome btw" }, { "id": 1392984, "postDate": "2021-07-19T09:25:31.183Z", "content": "Can you access imgur.com ? Images are at
\nhttps://i.imgur.com/hQqp1m7.png
\nhttps://i.imgur.com/Gz073Xc.png
", "rawMarkdown": "Can you access imgur.com ? Images are at\n\nhttps://i.imgur.com/hQqp1m7.png\n\nhttps://i.imgur.com/Gz073Xc.png", "votes": 1 }, { "id": 1392986, "postDate": "2021-07-19T09:31:41.110Z", "content": "I see! Thank you!
", "rawMarkdown": "I see! Thank you!" }, { "id": 1394259, "postDate": "2021-07-20T08:57:16.160Z", "content": "I can see it now, thx
", "rawMarkdown": "I can see it now, thx" } ] }, { "id": 1404964, "postDate": "2021-07-30T11:47:46.827Z", "content": "@cpmpml Hi :)
\nWould you mind sharing the code you used to run the above tests? I am keen to learn the way you did so I can apply this technique in future.
", "rawMarkdown": "@cpmpml Hi :)\n\nWould you mind sharing the code you used to run the above tests? I am keen to learn the way you did so I can apply this technique in future." }, { "id": 1393884, "postDate": "2021-07-20T02:17:31.293Z", "content": "Yes. I ran my previous xgboost code, but it doesn't work well.
\nIt looks like my computer will burn in summer…
Agreed! I ran a Kolmogorov-Smirnov / Malmgren-Svensson / Cramér-von Mises, and they are now all within error bars. Before, it would look like the \"wild west\"
", "votes": 1, "replies": [] }, { "id": 1391420, "author_name": "SleepMaster", "author_url": "", "post_date": "2021-07-17T15:20:22.443000", "content": "sorry why I cannot see the figure?
\n\n", "votes": 1, "replies": [ { "id": 1391492, "author_name": "CPMP", "author_url": "", "post_date": "2021-07-17T16:42:24.363000", "content": "I reran all the EDA I did previously on the new data and I could not find the statistical differences I once found. I hope we're safe now.
\nYop picture is form original data, bottom one is from new data. x is mean of power for snippets 0,2,4 and y is mean of power for snippets 1,3,5. Power is the element wise squared image. Red is target 1.
\n\n
\n
Given mean is almost zero, power is almost the same as variance. We see that on new data variance is distributed the same way for each target values. It was not the case with original data.
\n
I can see it in your comment. What browser do you use? If you open the link to the image can you see it?
", "votes": 1, "replies": [] }, { "id": 1392971, "author_name": "ForcewithMe", "author_url": "", "post_date": "2021-07-19T09:12:50.727000", "content": "I cannot see it too. I use chrome btw
", "votes": 0, "replies": [] }, { "id": 1392984, "author_name": "CPMP", "author_url": "", "post_date": "2021-07-19T09:25:31.183000", "content": "Can you access imgur.com ? Images are at
\nhttps://i.imgur.com/hQqp1m7.png
\nhttps://i.imgur.com/Gz073Xc.png
", "votes": 1, "replies": [] }, { "id": 1392986, "author_name": "ForcewithMe", "author_url": "", "post_date": "2021-07-19T09:31:41.110000", "content": "I see! Thank you!
", "votes": 0, "replies": [] }, { "id": 1394259, "author_name": "SleepMaster", "author_url": "", "post_date": "2021-07-20T08:57:16.160000", "content": "I can see it now, thx
", "votes": 0, "replies": [] } ] }, { "id": 1404964, "author_name": "gao-hongnan", "author_url": "", "post_date": "2021-07-30T11:47:46.827000", "content": "@cpmpml Hi :)
\nWould you mind sharing the code you used to run the above tests? I am keen to learn the way you did so I can apply this technique in future.
", "votes": 0, "replies": [] }, { "id": 1393884, "author_name": "WOOSUNG YOON", "author_url": "", "post_date": "2021-07-20T02:17:31.293000", "content": "Yes. I ran my previous xgboost code, but it doesn't work well.
\nIt looks like my computer will burn in summer…