IMPROVING CLASSIFICATION ACCURACY IN CROWD SOURCING WITH RANDOM FOREST ALGORITHM

Crowd sourcing involves collection of work information or opinion from a large group of people, who submit the data as comments or reviews. The increasing popularity of crowd sourcing market enables the application of the classification tasks. The methods for traditional classification tasks are Dynamic Label Acquisition and Answer Aggregation. The existing works do not exploit the Label inference and acquisition phase, which results inability of making a proper budget allocation for labels. In addition, label mismatch and multi-label tasks are some other problemsfaced in the existing works. To eradicate these problems, it is proposed to use Random Forest Algorithm (RFA) for crowd sourcing classification. The objective of this work is to improve the efficiency in time series by using a Dynamic Resource algorithm, makes a best analysis for classification in crowd sourcing and improves the classification accuracy in crowd sourcing applications. RFA is an ensemble learning method, which operates by constructing a multitude of decision tree at training time and results with the classes. RFA applies a technique of Bootstrap Aggregation to produce the final result with high accuracy.