![\"Test_acoustic\"](\"https://i.imgur.com/KlnPi5D.png\")
So. I plotted acoustic_data from test sequentially from the zipped folder. Notice that the min and max values do not appear to match experiment p4677's min and max according to the charts below. Makes me wonder 🤔.
I counted 9 full EQs in p4677. Based on my plot of test, it is not clear how many EQ's there are. However, if you take use all peaks of greater than 3000 as an EQ, then there are approximately 9 EQs in test based on my plot (remember we do not know the order).
\n\nFurthermore, we can see that training acoustics exceed the 3000 upper and lower bounds as seen on the charts of exp. p4677.
\n\nedit: so upon further reading, there are multiple loads recorded in p4677. which explains the mismatch in acoustic_data, however, it is safe to assume that the data is p4677 since Bertrand said, \n> The time series is stationary and the physics unchanged except for a small experimental artifact which is that some material from the fault is progressively lost as the experiment proceeds.
According to the paper where I pulled the image from, \n> In the first experiment we analyze, the normal load is fixed at 2.5 MPa. In the second experiment, that we analyze in the next sections, the normal load is constant at plateaus of 4, 5, 6, and 7 MPa.
\n\nThis can also explain why acoustic data is in \"arb. units\"
\n\n
![\"p4677\"](\"https://i.imgur.com/LUio4n1.png\")
![\"full_train\"](\"https://i.imgur.com/YznPIk6.png\")
So. I plotted acoustic_data from test sequentially from the zipped folder. Notice that the min and max values do not appear to match experiment p4677's min and max according to the charts below. Makes me wonder 🤔.
I counted 9 full EQs in p4677. Based on my plot of test, it is not clear how many EQ's there are. However, if you take use all peaks of greater than 3000 as an EQ, then there are approximately 9 EQs in test based on my plot (remember we do not know the order).
\n\nFurthermore, we can see that training acoustics exceed the 3000 upper and lower bounds as seen on the charts of exp. p4677.
\n\nedit: so upon further reading, there are multiple loads recorded in p4677. which explains the mismatch in acoustic_data, however, it is safe to assume that the data is p4677 since Bertrand said, \n> The time series is stationary and the physics unchanged except for a small experimental artifact which is that some material from the fault is progressively lost as the experiment proceeds.
According to the paper where I pulled the image from, \n> In the first experiment we analyze, the normal load is fixed at 2.5 MPa. In the second experiment, that we analyze in the next sections, the normal load is constant at plateaus of 4, 5, 6, and 7 MPa.
\n\nThis can also explain why acoustic data is in \"arb. units\"
\n\n
What is new compared to https://www.kaggle.com/c/LANL-Earthquake-Prediction/discussion/90664 ?
", "rawMarkdown": "What is new compared to https://www.kaggle.com/c/LANL-Earthquake-Prediction/discussion/90664 ?", "votes": null }, { "id": "539780", "postDate": "05/30/2019 14:07:54", "content": "@cpmpml not much. I had originally published the topic prematurely as a statement that the train and test from our competition were not from p4677. Then as I kept reading into it, I decided I was wrong so I edited the topic to state that it is from p4677. Basically I was wrong. I don't want to re-write or delete anything however. There are basically multiple loads tested in p4677, which made me confused, but I'm now certain that the test data is from p4677 at the fixed 2.5 MPa. Thanks for asking for clarification.
", "rawMarkdown": "cpmpml not much. I had originally published the topic prematurely as a statement that the train and test from our competition were not from p4677. Then as I kept reading into it, I decided I was wrong so I edited the topic to state that it is from p4677. Basically I was wrong. I don't want to re-write or delete anything however. There are basically multiple loads tested in p4677, which made me confused, but I'm now certain that the test data is from p4677 at the fixed 2.5 MPa. Thanks for asking for clarification.", "votes": null }, { "id": "539811", "postDate": "05/30/2019 14:52:13", "content": "Thanks for the answer. I was afraid I missed something ;)
", "rawMarkdown": "Thanks for the answer. I was afraid I missed something ;)", "votes": null } ], "comments": [ { "id": 539758, "author_name": "cpmpml", "author_url": "", "post_date": "05/30/2019 13:42:55", "content": "What is new compared to https://www.kaggle.com/c/LANL-Earthquake-Prediction/discussion/90664 ?
", "votes": null, "replies": [ { "id": 539780, "author_name": "teeyee314", "author_url": "", "post_date": "05/30/2019 14:07:54", "content": "@cpmpml not much. I had originally published the topic prematurely as a statement that the train and test from our competition were not from p4677. Then as I kept reading into it, I decided I was wrong so I edited the topic to state that it is from p4677. Basically I was wrong. I don't want to re-write or delete anything however. There are basically multiple loads tested in p4677, which made me confused, but I'm now certain that the test data is from p4677 at the fixed 2.5 MPa. Thanks for asking for clarification.
", "votes": null, "replies": [] }, { "id": 539811, "author_name": "cpmpml", "author_url": "", "post_date": "05/30/2019 14:52:13", "content": "Thanks for the answer. I was afraid I missed something ;)
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "535608": "So. I plotted `acoustic_data` from test sequentially from the zipped folder. Notice that the min and max values do not appear to match experiment p4677's min and max according to the charts below. Makes me wonder 🤔.\n\nI counted 9 full EQs in p4677. Based on my plot of test, it is not clear how many EQ's there are. However, if you take use all peaks of greater than 3000 as an EQ, then there are approximately 9 EQs in test based on my plot (remember we do not know the order).\n\nFurthermore, we can see that training acoustics exceed the 3000 upper and lower bounds as seen on the charts of exp. p4677.\n\nedit: so upon further reading, there are multiple loads recorded in p4677. which explains the mismatch in `acoustic_data`, however, it is safe to assume that the data is p4677 since Bertrand said, \n> The time series is stationary and the physics unchanged except for a small experimental artifact which is that some material from the fault is progressively lost as the experiment proceeds.\n\nAccording to the paper where I [pulled the image from](https://permalink.lanl.gov/object/tr?what=info%3Alanl-repo%2Flareport%2FLA-UR-17-29312), \n> In the first experiment we analyze, the normal load is fixed at 2.5 MPa. In the second experiment, that we analyze in the next sections, the normal load is constant at plateaus of 4, 5, 6, and 7 MPa.\n\nThis can also explain why acoustic data is in \"arb. units\"\n\n\n\n\n\n", "539758": "What is new compared to https://www.kaggle.com/c/LANL-Earthquake-Prediction/discussion/90664 ?", "539780": "cpmpml not much. I had originally published the topic prematurely as a statement that the train and test from our competition were not from p4677. Then as I kept reading into it, I decided I was wrong so I edited the topic to state that it is from p4677. Basically I was wrong. I don't want to re-write or delete anything however. There are basically multiple loads tested in p4677, which made me confused, but I'm now certain that the test data is from p4677 at the fixed 2.5 MPa. Thanks for asking for clarification.", "539811": "Thanks for the answer. I was afraid I missed something ;)" }, "source": "meta" }