In practice, we often encounter some missing and outlier data of a Building Integrated Photovoltaic system (BIPVs) among the data acquisitions. It may because of computer shut down, the electric pulse to sensitive sensor, thus affect the reliability of the whole data. Without complete data, we cannot catch the whole picture of the annual electric power generated respect to the BIPV system. The paper constructed a regression model according to the properties of electricity generated theory of photovoltaic, and related it to the characteristic of the data collection. Using the regression model and data collected from a PVs installed on the roof of a house located at Kao-Hsiung, this paper showed how the missing and outlier data can be interpolated properly. By using the seasonal model, the regression analysis showed that all variables in seasonal model are all significance to reject the hypothesis of no effect on the electric yield of the PVs (H0: the coefficient = 0) at 5% significant level. Finally, using the estimates coefficient of the estimators the missing and outlier data were interpolated, the daily yields of the BIPV system of the whole year were plotted.