Higher output or yield of cereals especially maize is very important to the economy of Ghana. This largely because several households in Ghana rely on foods which are prepared from maize. The backward and forward linkages that maize has with other facets of the economy such as poultry, manufacturing, fishery among others also makes it very eminent to boost the production of maize across the country. The negative ramifications of climate change on cereal production are also becoming very evident in Ghana. This particular study relied on extended vector autoregression model to predict the effect of climate data and maize yield data which were gathered over a period of 30 years from the northern sector of Ghana where maize is predominantly cultivated. Regarding the two performance evaluation metrics, the LSTM achieved 92.67% average accuracy performance across six trails. This performance was 3.17% higher than other models such as deep neural network model. Again, the LSTM achieved 97.17% prediction precision which was higher than other models such as deep neural network model. The models which were used in this study were efficient and robust since they were all able to achieve accuracy and precision performance metrics score of above 90%. The multiple R-square value of 0.749744 within the extended vector autogestion model showed that rainfall predicted 74.97% variance in maize yield over the period of analysis. Again, the observed multiple R-square value of 0.788119 also shows that temperature had 78.81% predictive effect on maize yield over the period of analysis. The study concluded that
temperature serve as the highest climatic element with relatively higher predictive effect on maize yield at Northern sector.