In this paper, we describe the formatting guidelines for ACM SIG Proceedings. Knowledge discovery is the process of analysing data for future planning. The nature of data and anomalies are different in different record data sets. The problem of detecting contextual anomalies in data sets is also different from the traditional anomaly detection problem. A time series is a chronological sequence of observations on a particular variable. Usually the observations are taken at regular intervals (days, months, years), but the sampling could be irregular. A time series analysis consists of two steps- building a model that represents a time series, and using the model to predict (forecast) future values. Anomaly depends on many factors such as temperature (season), Soil type etc. In our paper we proposed semi supervised learning for sericulture database. In sericulture the production of cocoons are analysed. The best, average and poor class label values depend on two approaches (1) Attributes like process knowledge, temperature, soil type, variety of mulberry plantation, use of fertilizers, turn of plantation etc. (2) Attribute contribution in terms of percentage. In semi supervised learning method in which input is line data stream and huge in nature. In the data stream, we can define class label val0ues for some instances but few instances are outlier. Outlier instances can be analysed using unsupervised learning. For semi supervised learning we are implementing Bayesian classification and rule based algorithm.

data set KDS sericulture