Using a 15-year (1995 to 2009) climatology of 1500 UTC warm-season (May through September) rawinsonde observation (RAOB) data from the Cape Canaveral Air Force Station (CCAFS) Skid Strip (KXMR) and 5 minute wind data from 36 wind towers on CCAFS and Kennedy Space Center (KSC), several convective wind forecasting techniques currently employed by the 45th Weather Squadron (45 WS) were evaluated. Present forecasting methods under evaluation include examining the vertical equivalent potential temperature (θe) profile, vertical profiles of wind spend and direction, and several wet downburst forecasting indices. Although previous research found that currently used wet downburst forecasting methods showed little promise for forecasting convective winds, it was carried out with a very small sample, limiting the reliability of the results. Evaluation versus a larger 15-year dataset was performed to truly assess the forecasting utility of these methods in the central Florida warm-season convective environment. In addition, several new predictive analytic based forecast methods for predicting the occurrence of warm-season convection and its associated wind gusts were developed and validated. This research was performed in order to help the 45 WS better forecast not only which days are more likely to produce convective wind gusts, but also to better predict which days are more likely to yield warning criteria wind events of 35 knots or greater, should convection be forecasted. Convective wind forecasting is a very challenging problem that requires new statistically based modeling techniques since conventional meteorologically based methods do not perform well. New predictive analytic based forecasting methods were constructed using R statistical software and incorporate several techniques including multiple linear regression, logistic regression, multinomial logistic regression, classification and regression trees (CART), and ensemble CART using bootstrapping. All of these techniques except the ensemble CART methods were built with data from the 1995 to 2007 warm-seasons and validated with a separate independent dataset from the 2008 and 2009 warm-seasons. Ensemble CART models were built using randomly selected data from the 1995 to 2009 RAOB dataset and validated with data not used in constructing the models. Three different ensemble CART algorithms including the random forests, bagging, and boosting algorithms were tested to find the best performing model. Quantitative verification results suggest that the presently used convection and wet downburst forecasting techniques do not show much operational promise. As such, it is not recommended that the 45 WS use vertical profiles of θe, wind speed, or wind direction to make specific predictions for which days are likely to produce convection or warning threshold wind gusts. None of the wet downburst indices used displayed much potential either. Although, the linear regression based predictive analytic models do not perform too well, CART based models perform better, especially those that utilize a binary response variable. Of the new techniques, the ensemble CART models displayed the most promise with the boosting algorithm showing nearly perfect results for predicting which days would produce convection and which days would produce warning threshold winds should convection be predicted.