{ "id": 68348, "title": "Google Store Data & public kernel with leak data", "url": "/competitions/ga-customer-revenue-prediction/discussion/68348", "author_name": "", "post_date": "2018-10-11T19:42:44.843922100Z", "votes": 21, "comment_count": 7, "views": 0, "content": "

Hi guys,\nI have downloaded and uploaded the google analytics data https://www.kaggle.com/satian/exported-google-analytics-data
\nThe data is for train and test date for users who have done transactions. I am also trying to upload a kernel with the notebook that helped me in getting the 2nd position.
\nI know the leak is stupid but I am really interested in knowing how much someone can exploit the leak.
\nYou can simply merge the datasets as the client id after '.' is visitId.
\nI will update the thread once my script gets updated but it is very basic code and is using Olivier's sum of predictions kernel.
\nI don't know what the future holds for this competition but like I said really interested in knowing how much the leak can be exploited.
\nI am just a beginner and no way i would have scored this high without the leak, hence I am sharing what I have done.
\nEDIT: This is the kernel https://www.kaggle.com/satian/story-of-a-leak?scriptVersionId=6422532

", "messages": [ { "id": "402491", "postDate": "10/11/2018 19:42:44", "content": "

Hi guys,\nI have downloaded and uploaded the google analytics data https://www.kaggle.com/satian/exported-google-analytics-data
\nThe data is for train and test date for users who have done transactions. I am also trying to upload a kernel with the notebook that helped me in getting the 2nd position.
\nI know the leak is stupid but I am really interested in knowing how much someone can exploit the leak.
\nYou can simply merge the datasets as the client id after '.' is visitId.
\nI will update the thread once my script gets updated but it is very basic code and is using Olivier's sum of predictions kernel.
\nI don't know what the future holds for this competition but like I said really interested in knowing how much the leak can be exploited.
\nI am just a beginner and no way i would have scored this high without the leak, hence I am sharing what I have done.
\nEDIT: This is the kernel https://www.kaggle.com/satian/story-of-a-leak?scriptVersionId=6422532

", "rawMarkdown": "Hi guys,\nI have downloaded and uploaded the google analytics data https://www.kaggle.com/satian/exported-google-analytics-data
\nThe data is for train and test date for users who have done transactions. I am also trying to upload a kernel with the notebook that helped me in getting the 2nd position.
\nI know the leak is stupid but I am really interested in knowing how much someone can exploit the leak.
\nYou can simply merge the datasets as the client id after '.' is visitId.
\nI will update the thread once my script gets updated but it is very basic code and is using Olivier's sum of predictions kernel.
\nI don't know what the future holds for this competition but like I said really interested in knowing how much the leak can be exploited.
\nI am just a beginner and no way i would have scored this high without the leak, hence I am sharing what I have done.
\nEDIT: This is the kernel https://www.kaggle.com/satian/story-of-a-leak?scriptVersionId=6422532", "votes": null }, { "id": "402566", "postDate": "10/11/2018 22:40:38", "content": "

The client id after '.' is, I believe, the first visitID for a given visitor.\nKnowing that, you can potentially exploit the leak more.

\n\n

However, some of the sessions on GA are not in our data. Therefore this might lead to mistakes with purchasing visitors. But maybe not with non purchasing visitors.

\n\n

For purchasing visitors, we can look at the session detail by clicking on the visitID on GA. Then you can connect the dots with pageviews/hits/etc.

", "rawMarkdown": "The client id after '.' is, I believe, the first visitID for a given visitor.\nKnowing that, you can potentially exploit the leak more.\n\nHowever, some of the sessions on GA are not in our data. Therefore this might lead to mistakes with purchasing visitors. But maybe not with non purchasing visitors.\n\nFor purchasing visitors, we can look at the session detail by clicking on the visitID on GA. Then you can connect the dots with pageviews/hits/etc.", "votes": null }, { "id": "402616", "postDate": "10/12/2018 02:06:12", "content": "

Thank you for sharing the external data

", "rawMarkdown": "Thank you for sharing the external data", "votes": null }, { "id": "402919", "postDate": "10/12/2018 15:17:22", "content": "

Thank you for the kernel ! Very useful.

\n\n

Did you do any tuning on LGBM or you just took some \"standard\" parameters ?

", "rawMarkdown": "Thank you for the kernel ! Very useful.\n\nDid you do any tuning on LGBM or you just took some \"standard\" parameters ?", "votes": null }, { "id": "402966", "postDate": "10/12/2018 16:38:14", "content": "

I just used public kernel by Olivier and simply fit the data.

", "rawMarkdown": "I just used public kernel by Olivier and simply fit the data.", "votes": null }, { "id": "403221", "postDate": "10/13/2018 05:20:48", "content": "

Very Powerful!!!

", "rawMarkdown": "Very Powerful!!!", "votes": null }, { "id": "404644", "postDate": "10/16/2018 05:49:47", "content": "

Thanks Ankit for posting this, this is immensely helpful for beginners, such as myself.

", "rawMarkdown": "Thanks Ankit for posting this, this is immensely helpful for beginners, such as myself.", "votes": null }, { "id": "452891", "postDate": "01/09/2019 10:13:05", "content": "

does someone know how to extract data from Gstore or my own Google analytics on visitorid/fullvisitorid level? any tips?

", "rawMarkdown": "does someone know how to extract data from Gstore or my own Google analytics on visitorid/fullvisitorid level? any tips?", "votes": null } ], "comments": [ { "id": 402566, "author_name": "sdoria", "author_url": "", "post_date": "10/11/2018 22:40:38", "content": "

The client id after '.' is, I believe, the first visitID for a given visitor.\nKnowing that, you can potentially exploit the leak more.

\n\n

However, some of the sessions on GA are not in our data. Therefore this might lead to mistakes with purchasing visitors. But maybe not with non purchasing visitors.

\n\n

For purchasing visitors, we can look at the session detail by clicking on the visitID on GA. Then you can connect the dots with pageviews/hits/etc.

", "votes": null, "replies": [] }, { "id": 402616, "author_name": "nyleve", "author_url": "", "post_date": "10/12/2018 02:06:12", "content": "

Thank you for sharing the external data

", "votes": null, "replies": [] }, { "id": 402919, "author_name": "xtelima", "author_url": "", "post_date": "10/12/2018 15:17:22", "content": "

Thank you for the kernel ! Very useful.

\n\n

Did you do any tuning on LGBM or you just took some \"standard\" parameters ?

", "votes": null, "replies": [ { "id": 402966, "author_name": "satian", "author_url": "", "post_date": "10/12/2018 16:38:14", "content": "

I just used public kernel by Olivier and simply fit the data.

", "votes": null, "replies": [] } ] }, { "id": 403221, "author_name": "kangzhang", "author_url": "", "post_date": "10/13/2018 05:20:48", "content": "

Very Powerful!!!

", "votes": null, "replies": [] }, { "id": 404644, "author_name": "shoubhik", "author_url": "", "post_date": "10/16/2018 05:49:47", "content": "

Thanks Ankit for posting this, this is immensely helpful for beginners, such as myself.

", "votes": null, "replies": [] }, { "id": 452891, "author_name": "", "author_url": "", "post_date": "01/09/2019 10:13:05", "content": "

does someone know how to extract data from Gstore or my own Google analytics on visitorid/fullvisitorid level? any tips?

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "402491": "Hi guys,\nI have downloaded and uploaded the google analytics data https://www.kaggle.com/satian/exported-google-analytics-data
\nThe data is for train and test date for users who have done transactions. I am also trying to upload a kernel with the notebook that helped me in getting the 2nd position.
\nI know the leak is stupid but I am really interested in knowing how much someone can exploit the leak.
\nYou can simply merge the datasets as the client id after '.' is visitId.
\nI will update the thread once my script gets updated but it is very basic code and is using Olivier's sum of predictions kernel.
\nI don't know what the future holds for this competition but like I said really interested in knowing how much the leak can be exploited.
\nI am just a beginner and no way i would have scored this high without the leak, hence I am sharing what I have done.
\nEDIT: This is the kernel https://www.kaggle.com/satian/story-of-a-leak?scriptVersionId=6422532", "402566": "The client id after '.' is, I believe, the first visitID for a given visitor.\nKnowing that, you can potentially exploit the leak more.\n\nHowever, some of the sessions on GA are not in our data. Therefore this might lead to mistakes with purchasing visitors. But maybe not with non purchasing visitors.\n\nFor purchasing visitors, we can look at the session detail by clicking on the visitID on GA. Then you can connect the dots with pageviews/hits/etc.", "402616": "Thank you for sharing the external data", "402919": "Thank you for the kernel ! Very useful.\n\nDid you do any tuning on LGBM or you just took some \"standard\" parameters ?", "402966": "I just used public kernel by Olivier and simply fit the data.", "403221": "Very Powerful!!!", "404644": "Thanks Ankit for posting this, this is immensely helpful for beginners, such as myself.", "452891": "does someone know how to extract data from Gstore or my own Google analytics on visitorid/fullvisitorid level? any tips?" }, "source": "meta" }