Safety analytics is a data-driven approach with an aim to help in enhancing safety in a workplace. One of the major potential areas of application of safety analytics lies in the forecasting of the occurrence of incidents in a workplace. In this study, on a 21-month-long incident investigation process safety dataset in a steel plant, an integration of time series, and machine learning approach is implemented. Firstly, the elements of an incident path are extracted from the brief description of the incidents from the dataset. Then, the number of occurrences of incidents in a month is forecasted through the development of the Autoregressive Integrated Moving Average (ARIMA) model. The time interval of the event of a specific incident is obtained by utilizing a Long Short-Term Memory (LSTM) model. Finally, the development of a Random Forest model enabled the prediction of the day and hour of occurrence of an incident. The validation of the machine learning models was achieved using their predictions on test data as applicable, while the time series model forecasts the number of incidents for the immediate future. Thus the integration of the three models gives a holistic framework for incident forecasting, which helps in formulating targets and using resources for implementation of interventions.