This study presents a methodology using multivariate analysis: Principal Component Analysis (PCA) and Cluster Analysis (CA) to analyze data of hourly averaged speed in hours from 28 stations distributed in four states of Northeastern Brazil: Ceara with 10 stations, Paraiba with 5 stations, Pernambuco with 8 stations and Rio Grande do Norte with 5 stations. All stations are well distributed spatially and period of data between 1977 to 1981. The results of the Principal Component Analysis (PCA) showed that the coastal and mountainous regions have the greatest potential for energy generation results, in particularly at the stations of Acarau-CE and Macau-RN, while Barbalha-CE had the lowest potential, possibly due to its location. The Cluster Analysis (CA), using the Ward method, allowed the distribution of the stations into six homogeneous groups.