Smart tourists cannot be separated with mobile technology. With the gadget, tourist can find information about the destination, or supporting information like transportation, hotel, weather and exchange rate. They need prediction of traveling and visiting time, to arrange their journey. If traveling time has predicted accurately by Google Map using the location feature, visiting time has another issue. Until today, Google detects the user's position based on crowdsourcing data from customer visits to a specific location over the last several weeks. It cannot be denied that this method will give a valid information for the tourists. However, because it needs a lot of data, there are many destinations that have no information about visiting time. From the case study that we used, there are 626 destinations in East Java, Indonesia, and from that amount only 224 destinations or 35.78% has the visiting time. To complete the information and help tourists, this research developed the prediction model for visiting time. For the first data is tested statistically to make sure the model development was using the right method. Multiple linear regression become the common model, because there are six factors that influenced the visiting time, i.e. access, government, rating, number of reviews, number of pictures, and other information. Those factors become the independent variables to predict dependent variable or visiting time. From normality test as the linear regression requirement, the significant value was less than p that means the data cannot pass the statistic test, even though we transformed the data based on the skewness. Because of three of them are ordinal data and the others are interval data, we tried to exclude and include the ordinal by transform it to interval. We also used the Ordinal Logistic Regression by transform the interval data in dependent variable into ordinal data using Expectation Maximization, one of clustering algorithm in machine learning, but the model still did not fit even though we used 5 functions. Then we used the classification algorithm in machine learning by using 5 top algorithm which are Linear Regression, k-Nearest Neighbors, Decision Tree, Support Vector Machines, and Multi-Layer Perceptron. Based on maximum correlation coefficient and minimum root mean square error, Linear Regression with 6 independent variables has the best result with the correlation coefficient 20.41% and root mean square error 48.46%. We also compared with model using 3 independent variable, the best algorithm was still the same but with less performance. Then, the model was loaded to predict the visiting time for other 402 destinations.