### here sparse data is categorical data, not necessary we have to modify the code to do classification, and for multiclass by jinluyang Data Format =========== The input of this GBDT solver consists of a label vector (y), a dense matrix (XD), and a binary sparse matrix (XS). The input format of these two matrices are introduced in the following two sections. Dense Matrix ------------ The input format is: <label> <value_1> <value_2> ... . . . Note that to represent a dense matrix, we do not have to give indices. For example, 1 32 91 27 44 0 13 25 55 83 0 32 11 78 99 represents: y XD 1 32 91 27 44 0 13 25 55 83 0 32 11 78 99 Binary Sparse Matrix -------------------- The input format is: <label> <index_1> <index_2> ... . . . To represent a binary sparse matrix, we only need to know where non-zero elements are, so values are not specified. For example, 1 2 9 5 0 1 3 7 0 4 8 2 represents: y XS 1 0 1 0 0 1 0 0 0 1 0 1 0 1 0 0 0 1 0 0 0 0 1 0 1 0 0 0 1 0 Note that the labels in binary sparse matrix are just dummies. They do not have pratical use; please specify correct labels in dense matrix.