As we all now, the fullscreen column doesn't give us much information
After I realised people used the maximum for screen_coordinate_x and screen_coordinate_y to make up for this missing column, I decided to make this discussion explaining why it isn't a trust worthy approach and how you can improve it with a much more accurate value.
Taking the max for the x axis and the y axis makes two assumptions:
\nTo tackle the first problem, you need to find a room or fqid where the screen is locked (you can't change the camera position by moving) and where the player had to click on a very small item.
\nAfter finding it, you need to plot the screen clicks coordinates for that specific item.
\nIt will give you something like this:
\n ![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F12870466%2F260355ea3b926e9d8443e905d08ce8dd%2Ffor%20chris%20discussion.png?generation=1679288328771609&alt=media\")
\nAnd now, you know all the players that kept the standard resolution (the ones in the very dense cluster), and those who reduced/increased their window size by simply scaling the x and y values with the room coordinate that click represent !
To tackle the second problem, you can do this for multiple rooms along the game and you will be able to compare the results !
", "messages": [ { "id": 2188907, "postDate": "2023-03-20T05:03:38.747Z", "content": "As we all now, the fullscreen column doesn't give us much information
After I realised people used the maximum for screen_coordinate_x and screen_coordinate_y to make up for this missing column, I decided to make this discussion explaining why it isn't a trust worthy approach and how you can improve it with a much more accurate value.
Taking the max for the x axis and the y axis makes two assumptions:
\nTo tackle the first problem, you need to find a room or fqid where the screen is locked (you can't change the camera position by moving) and where the player had to click on a very small item.
\nAfter finding it, you need to plot the screen clicks coordinates for that specific item.
\nIt will give you something like this:
\n
\nAnd now, you know all the players that kept the standard resolution (the ones in the very dense cluster), and those who reduced/increased their window size by simply scaling the x and y values with the room coordinate that click represent !
To tackle the second problem, you can do this for multiple rooms along the game and you will be able to compare the results !
", "rawMarkdown": "**As we all now, the `fullscreen` column doesn't give us much information**\n\nAfter I realised people used the maximum for `screen_coordinate_x` and `screen_coordinate_y` to make up for this missing column, I decided to make this discussion explaining **why it isn't a trust worthy approach and how you can improve it with a much more accurate value.**\n\nTaking the max for the x axis and the y axis makes two assumptions: \n- **The player clicked on the far edges of his screen for both the x and y axis:** Playing the game, I wanted to see if their was any task that requires us to click at the top of our screen, and their is none, so unless the player willingly clicked at the top, we can't know his actual y coordiate.\n- **The player didn't change his screen resolution along the game**\n\n*To tackle the first problem*, you need to find a room or fqid where the screen is locked *(you can't change the camera position by moving)* and where the player had to click on a very small item.\nAfter finding it, you need to plot the **screen** clicks coordinates for that specific item.\nIt will give you something like this:\n \nAnd now, you know all the players that kept the standard resolution *(the ones in the very dense cluster)*, and those who reduced/increased their window size by simply scaling the x and y values with the room coordinate that click represent !\n\n*To tackle the second problem*, you can do this for multiple rooms along the game and you will be able to compare the results !", "votes": 5 } ], "comments": [], "raw_markdown_by_id": { "2188907": "**As we all now, the `fullscreen` column doesn't give us much information**\n\nAfter I realised people used the maximum for `screen_coordinate_x` and `screen_coordinate_y` to make up for this missing column, I decided to make this discussion explaining **why it isn't a trust worthy approach and how you can improve it with a much more accurate value.**\n\nTaking the max for the x axis and the y axis makes two assumptions: \n- **The player clicked on the far edges of his screen for both the x and y axis:** Playing the game, I wanted to see if their was any task that requires us to click at the top of our screen, and their is none, so unless the player willingly clicked at the top, we can't know his actual y coordiate.\n- **The player didn't change his screen resolution along the game**\n\n*To tackle the first problem*, you need to find a room or fqid where the screen is locked *(you can't change the camera position by moving)* and where the player had to click on a very small item.\nAfter finding it, you need to plot the **screen** clicks coordinates for that specific item.\nIt will give you something like this:\n \nAnd now, you know all the players that kept the standard resolution *(the ones in the very dense cluster)*, and those who reduced/increased their window size by simply scaling the x and y values with the room coordinate that click represent !\n\n*To tackle the second problem*, you can do this for multiple rooms along the game and you will be able to compare the results !" } }