This thesis addresses tool-life forecasting and tool-change decision support for an industrial tapping process in the metal-working sector. The study is conducted on a single production asset. The industrial goal is to improve tool management by estimating the remaining useful life (RUL) of each tool and supporting timely replacements to reduce nonconforming production and scrap while avoiding unnecessary downtime and premature tool changes. To achieve this, heterogeneous machine logs are integrated, including position level production and quality counters and machine-level sensor measurements such as temperatures, alarm/event information, and recorded tool-change events. A dataset engineering pipeline is developed to align these sources consistently over time and to construct cycle-based RUL targets and informative degradation features.