Decision Key-Value Feature Construction for Multihoming Big Data Network

The random forest algorithm under the MapReduce framework has too many redundant and irrelevant features, low training feature information, parallelization efficiency when dealing with multihoming big data network problems, so parallelism is based on information theory, norms proposed for (PRFITN). In this paper, technique used first builds a hybrid dimensional reduction approach (DRIGFN) focused gain Frobenius norm, successfully reducing number of features; then, an theory offered. This results in dimensionality-reduced dataset. Finally, suggested Reduce stage. features are grouped FGSIT strategy, stratified sampling employed to assure quantity building decision tree forest. When datasets provided as key/value pairs, it common want aggregate statistics across all objects same key. To acquire global classification achieve rapid equal distribution key-value pair redistribution method (RSKP) used, which improves cluster’s parallel efficiency. provides superior impact large networks, particularly numerous characteristics, according experimental findings. We can utilize selection extraction together. addition minimizing overfitting redundancy, lowering dimensionality contributes improved human interpretation cheaper computing costs through model simplicity.