A Parameter Estimation Model of Big Data Setup Based on Sampling Technique

Abstract

At the present time, we can see around us everywhere big data is available like health sectors, insurance, share market, government organizations, social networking. Basically, big data refers to a huge amount of datasets and that dataset can be stored in the form of structured or unstructured format. Those datasets are producing in every second via a wide range of sources such as Facebook and Twitter. The big data has a basic three pillars- Volume (size), Variety (types) and Velocity (speed). Big data is a prediction platform where lots of dataset are available in the form of raw material, so we need an appropriate technique to process it and predict some valuable information for the present as well as future use. The sampling techniques are applicable to handle large datasets where all data sets are not possible to process at a time. In this situation, the sampling techniques play a very important role in the processing of big huge data. In this paper, we have proposed a parameter estimation model for big data to predict the mean size of large data by using the sampling technique. The proposed model can reduce the processing time as compare to the existing model because the sampling mechanism is to select small sample (n) from the entire datasets (N) and estimate the unknown parameters which are known as prediction results. Furthermore, we have presented the demonstration of the proposed algorithm on dynamic dataset and also compared result with simple random sampling (SRS) technique.