Here, we include the data supporting the manuscript "Local electronic descriptors for solute-defect interactions in bcc refractory metals" authored by Y.J Hu, G. Zhao, B. Zhang, C. Yang, M. Zhang, Z.K. Liu, X. Qian, L. Qi. Under the "DFT-data" folder: 1) Folders named as "W-matrix" and "Ta-matrix". Under these two folders, there sub-folders named by the defect structures that have been stuided in the present work. In each of the "defect-named" subfolder: 1.1) One can excute a MatLab script named as "get_parameter.m" to calculate the electronic descriptors from the DFT-computed LDOSs. A EXCEL file named as "E-descriptors.xlsx" will be generated to include the calculated values of the electronic descriptors (.e.g, delta_dip, x_sp,...). 1.2) The DFT-computed electronic LDOS of each defect site can be found in the folder named as "DOS". Under the "DOS" folder, there is also a sub-folder named as "raw" where we included the VASP setup files (CONTCAR, KPOINTS, INCAR) and correspondingly calculated DOSCAR for the defect structure. 1.3) There is also a folder named as "E_int" where we included the VASP setup files for the calculations of solute-defect interaction energies for each defect site. Under "E_int", there are sub-folders named by numbers from "1,2,3,....". This number corresponds to the site index listed in Supplementary Tables S2 and S3 in the manuscript. The largest number corresponds to the reference configuration, where the solute atom is far away from the defect center. The DFT-computed total energy of the reference configuration is used as the reference to calculate the solute-defect interaction energy. 2) A folder named as "POTCARs for calculation". In this folder, we included the POTCAR files used for our DFT calculations. Under the "Regression code and data" folder: 1) The descriptors and reponses used for regressions and predicitions are summerized as .csv files. 2) There is a R script named as "linear_locfit_regression.R". By excuting the script (recommond using the Rstudio platform), the regression/prediction results will be generated into five .csv files (named as "X_Y_regression_results.csv" where X=Ta, W and Y=Hf, Ta, Re, Os, Pt) for the five stuided solute-matrix alloy pairs (i.e. W-Ta, W-Re, W-Pt, Ta-Hf and Ta-Os). 3) The format of the "X_Y_regression_results.csv" file: colume "system": Types of the defect structures colume "site_index": The number corresponds to the defect site index listed in Supplementary Tables S2 and S3 in the manuscript. colume "DFT_": The solute-defect interaction energies directly computed by DFT calculations(E_int_fix defined in the manuscript). colume "Linear_": The solute-defect interaction energies predicted from the two-descriptor Linear model (Eq. 1 in the manuscript). colume "Linear_f_rc_":The solute-defect interaction energies predicted from the Linear+f_r-c model (Eq. 4 in the manuscript).