The data represents Android apps that are labeled as benign or malicious. The goal is to build models that can identify malicious apps (or malware), so that they can be prevented from entering the market. Two sets of apps are provided. These datasets are csv files(AndroidAppsTrain.csv and AndroidAppsTest.csv). Each app in the train and test files is represented using 471 binary (0/1) features, which denote the presence/absence of various permissions, intent actions, discriminative APIs, obfuscation signatures, and native code signatures. For simplicity sake, we assume that the names of the features are f0, f2, · · · , f470. The last column of an instance/app (f471) corresponds to the class label y, which takes values 0 (benign) or 1 (malware).