Stream data mining is the process of excerpting knowledge structure from large, continuous data. For stream data, various techniques are proposed for preparing the data for data mining task. In recent years stream data have become a growing area for the researcher, but there are many issues occurring in classifying these data due to erroneous and noisy data. Change of trend in the data periodically produces major challenge for data miners. Column oriented data storage has shown fast access to data used in data mining. Various missing value replacement algorithms are implemented using MonetDB. This research also concentrates on incremental missing value replacement for stream data. The proposed method generates the value for the missing data considering the data type and data distribution. It also considers the concept drift in the data stream. The method will applied to different datasets and promising results will be expected