Predictive analytics is a methodology used to predict the outcome of future events with the use of historical data. Predictive analytics comes in very handy in various fields such as finance, manufacturing, healthcare, and even supply chain. Not only in those fields, but predictive analytics is also useful in managing inventory. However, we find that there is a lack of studies focusing on the implementation of predictive analytics to predict inventory status (overstock, understock) by considering inventory level and demand forecast. This study is inspired by a real-world problem at one of the largest FMCG companies in Indonesia. With so many product types to manage, this company often faces problems of understocked and overstocked inventory. This study attempts to solve that problem by employing big data and predictive analytics approaches. The gradient boosting model is used because it is an improvement of the decision tree model. The data that are used as predictors are inventory level, inventory week cover, historical sales, and demand forecast. The target variable for classification is inventory status which is divided into three classes, namely understock, normal, and overstock. Meanwhile, the target variable for the regression model is the amount of understock/overstock. The result of the classification model has an accuracy of 0.84 for category 1 products, 0.76 for category 2 products, and 0.74 for category 3 products. While the result of the regression model is an R2 of 0.89 for category 1 products, 0.76 for category 2 products, and 0.74 for category 3 products. The data that comes from the prediction model are visualized in a dashboard. The visualization dashboard displays the data using heatmaps and line graphs, so the information can be used for further analysis.