Uncertain data is the data that deviate from the exact and original values. Clustering the uncertain data is an imperative task in data mining since it provides unpredicted results. Many clustering techniques have been developed for processing the uncertain data but the robustness and accuracy were not increased. In order to increase the clustering accuracy, a Fully Recurrent Deep Neural Learning based X-means Data Clustering (FRDNL-XDC) technique is proposed for mining uncertain data with minimal time. In this technique, deep learning approach based X-means clustering is used to partition the whole dataset into different clusters for mining the uncertain data. The ‘x’ number of clusters and centroids are initialized randomly in the hidden layer. The bivariate correlation coefficient is used to find the correlation between the centroid of the cluster and the data to be clustered. Sigmoid activation function is used at the output layer where it maps the data with the centroid of the cluster to form certain data using Akaike information criterion with higher accuracy and minimum time. In FRDNL-XDC technique, feedback connections formulate to group the similar data with the minimal false positive rate. For validation, the proposed FRDNL-XDC technique is compared with the existing clustering techniques namely UKmeans clustering mechanism and UK-medoids-SMDM. The results obtained on El nino dataset show that the proposed FRDNL-XDC technique increases the clustering accuracy with less false positive rate and time complexity than the state-of-the-art methods. © 2005 – ongoing JATIT & LLS.