ABSTRACT Road traffic rapidity forecasting may be a testing problem done intelligent transport system (ITS) and need picked up expanding attentions. Existing meets expectations would principally In light of crude rapidity sensing data gotten from framework sensors or explored vehicles that are restricted Toward unreasonable cosset for sensor sending And upkeep. With meagre pace observations, accepted routines depended main on pace sensing data need aid insufficient, particularly the point when emergencies such as traffic mishaps happen. On location the problem, this paper plans on enhance those way traffic rapidity forecasting Toward fusing universal pace sensing data for new-type “sensing” data from cross area sources, for example, tweet sensors from Online networking and path sensors from guide And traffic administration platforms. Mutually displaying majority of the data starting with different datasets acquires huge numbers challenges, including area questionable matter of low-determination data, dialect vagueness of traffic portrayal in writings Also heterogeneity of cross-domain data. Because of the opposition on this disputes, we exhibit a bound together probabilistic system, known as TopicEnhanced Gaussian procedure amassed representation (TEGPAM), comprising about apparatus, i. E. Area disaggregation representation, traffic subject representation Also traffic rapidity Gaussian transform representation, that coordinate new-type data with customary data. Investigations looking into true data from two expansive urban areas On America accept the adequacy and effectiveness by our representation.