With the popularization of smartphones, the needs for location information, especially indoors, are rapidly growing these days. Location-based weather forecasting service and advertisement are well-known services that require user`s current location information. There are many techniques and approaches to provide indoor positioning information indoors. Among the techniques, the Wi-Fi based positioning technique has been widely used, and fingerprint-based localization is preferred because of its advantage in accuracy. However, the accuracy of localization gradually degrades as the Wi-Fi environment changes. In order to prevent a positioning system from the degrading, a radio map, which holds the Wi-Fi environment information, should be updated occasionally to reflect the Wi-Fi environment changes in the radio map. Recalibration is commonly used to update a radio map, but it usually requires a lot of time and effort. In this paper, we develop a method that can update a radio map automatically by using feedback data from numerous users. Various sensed data have been used to estimate collected locations of the fingerprints more accurately. To ensure correct updating of the radio map, we developed an optimization algorithm and a filtering method to handle erroneous data. An experiment was conducted to test the developed method at KAIST campus in Daejeon, Korea. Compared with having no update, the method showed an improvement in the accuracy of localization and achieved accuracy comparable to manual calibration in spite of using user feedback data. This indicates that our method can greatly contribute to reducing the radio map calibration efforts with only a minor sacrificing in positioning accuracy.