train[is.na(train)] <- 0 \ndoesn't work* but double negation
\n\ntrain[!(!(is.na(train)))] <- 0\nworks fine
\n\n*) error, Scientific Linux 7.2, R version 3.3.1, 160GB RAM, train=numeric+categorical
", "messages": [ { "id": "133271", "postDate": "09/01/2016 06:57:15", "content": "train[is.na(train)] <- 0 \ndoesn't work* but double negation
\n\ntrain[!(!(is.na(train)))] <- 0\nworks fine
\n\n*) error, Scientific Linux 7.2, R version 3.3.1, 160GB RAM, train=numeric+categorical
", "rawMarkdown": "train[is.na(train)] <- 0 \r\n\r\ndoesn't work* but double negation \r\n\r\n train[!(!(is.na(train)))] <- 0\r\n\r\nworks fine\r\n\r\n*) error, Scientific Linux 7.2, R version 3.3.1, 160GB RAM, train=numeric+categorical", "votes": null }, { "id": "133293", "postDate": "09/01/2016 11:11:26", "content": "Hi Grzegorz Sionkowski,
\n\nI can't understant why this occured.
\nDoes double negation have better characteristics ?
There is an error connected with the size of integer (size of list?) used in this operation. I suppose it is too small to determine the positions of all NA in the dataset but big enough to determine the positions of available values (first negation) . The second negation probably points to set 0 everywhere but the places where the values are available, not everywhere where the values are not available as in the first case. Of course, the same effect may be obtained in other ways (working with smaller datasets), but they are not so intriguing as the one described above.
", "rawMarkdown": "There is an error connected with the size of integer (size of list?) used in this operation. I suppose it is too small to determine the positions of all NA in the dataset but big enough to determine the positions of available values (first negation) . The second negation probably points to set 0 everywhere but the places where the values are available, not everywhere where the values are not available as in the first case. Of course, the same effect may be obtained in other ways (working with smaller datasets), but they are not so intriguing as the one described above.", "votes": null }, { "id": "133402", "postDate": "09/01/2016 21:57:33", "content": "First of all, a list is a vector so it can contain at most 2^32-1 number i.e. 4,294,967,296.A list is a vector
\n\nthe biggest table has less than 1,200,000 rows and also less than 2300 columns if all the cells are "na" the size would be 1200000* 2300 = 2,760,000,000 at most so it is not the case.\nI also dont knnow the right answer but it may be the maximum RAM space available to R no the machine.
", "rawMarkdown": "First of all, a list is a vector so it can contain at most 2^32-1 number i.e. 4,294,967,296.[A list is a vector][1]\r\n\r\nthe biggest table has less than 1,200,000 rows and also less than 2300 columns if all the cells are \"na\" the size would be 1200000* 2300 = 2,760,000,000 at most so it is not the case.\r\nI also dont knnow the right answer but it may be the maximum RAM space available to R no the machine.\r\n [1]: http://stackoverflow.com/questions/8594814/what-are-the-differences-between-r-vector-and-r-list-data-types", "votes": null } ], "comments": [ { "id": 133293, "author_name": "toshik", "author_url": "", "post_date": "09/01/2016 11:11:26", "content": "Hi Grzegorz Sionkowski,
\n\nI can't understant why this occured.
\nDoes double negation have better characteristics ?
There is an error connected with the size of integer (size of list?) used in this operation. I suppose it is too small to determine the positions of all NA in the dataset but big enough to determine the positions of available values (first negation) . The second negation probably points to set 0 everywhere but the places where the values are available, not everywhere where the values are not available as in the first case. Of course, the same effect may be obtained in other ways (working with smaller datasets), but they are not so intriguing as the one described above.
", "votes": null, "replies": [] }, { "id": 133402, "author_name": "afshinamiri", "author_url": "", "post_date": "09/01/2016 21:57:33", "content": "First of all, a list is a vector so it can contain at most 2^32-1 number i.e. 4,294,967,296.A list is a vector
\n\nthe biggest table has less than 1,200,000 rows and also less than 2300 columns if all the cells are "na" the size would be 1200000* 2300 = 2,760,000,000 at most so it is not the case.\nI also dont knnow the right answer but it may be the maximum RAM space available to R no the machine.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "133271": "train[is.na(train)] <- 0 \r\n\r\ndoesn't work* but double negation \r\n\r\n train[!(!(is.na(train)))] <- 0\r\n\r\nworks fine\r\n\r\n*) error, Scientific Linux 7.2, R version 3.3.1, 160GB RAM, train=numeric+categorical", "133293": "Hi Grzegorz Sionkowski,\r\n\r\nI can't understant why this occured.