Solar winds, geomagnetic storms cause significant disturbances in the upper layers of the upper atmosphere. These disturbances have adverse effects on radio waves and satellite communications. It plays an important role on the total electron emission by the great storm. The effects of solar winds can be observed even after geomagnetic activity has ceased. It takes a long time to affect the components of the ionosphere. In electron density measurement, when negative ionospheric storms are observed, the storm causes oxygen/Nitrogen depletion, triggering atmospheric disturbances. In this study, the effect of TEC (Total Electron Content) change in the ionosphere on the city of Ankara in Turkey was tried to be determined. It is concerned with the analysis and modeling of the temporal variation of the daily TEC in the ionosphere for the study region. Typically, TEC data is examined to observe and interpret the daily changes in data through graphical representation. However, in this study, it is not just about observation and interpretation; a predictive study is conducted on the data using the Wavelet analysis method and machine learning algorithms. Additionally, in the literature review conducted, no such study was found for the city of Ankara in the year 2017. In this regard, this study holds the distinction of being the first of its kind with the dataset specific to Ankara. The temporal variation of TEC data for the period January 1–December 31, 2017, has been examined and modeled. 60% of the one-year daily data was used for training, 20% for evaluation and 20% for testing. In the first part of the study, firstly, predictions were made with machine learning algorithms. In the second part of the study, wavelet transforms and large, medium and small-scale changes in TEC and precipitation data were examined. TEC data are modeled with support vector machine called linear regression, decision trees and machine learning based prediction methods. As a result, it has been determined that different algorithms show different success rates in the same geographical region. It has been determined that the structure of the datasets has a significant effect on the success rate of the algorithms. RMSE shows better modeling (0.02 rmse rate) for Linear Support Vector Machine (LSVM) modeling.