The main goal of hydraulic fracturing is to create a highly conductive fracture system that will improve the inflow performance and increase the ultimate reservoir recovery. Not properly designed process leads to underperformance of treated wells and can negatively impact the reservoir. Accurate hydraulic fracturing design is of great importance for post-job efficiency. As in many other areas, the improvements are a natural consequence of previous measurement and detailed analysis. Therefore, evaluating the historical frac jobs could help to improve the planning and increase execution efficiency. During this master's thesis writing, a practical tool for evaluating hydraulic fracturing performance is developed. The tool is based on a data-driven approach that helps in interpreting real-time data. Proposed algorithms automatically classify each timestamp of the treatment schedule and assign the stage label. Support vector machines and neural networks are used to classify the operation stage. These models are trained and evaluated on the datasets recorded on several wells. This thesis aims to set the metrics that could be generated based on the hydraulic fracturing job monitoring and provide valuable feedback about job efficiency. Defined metrics and a data-driven approach help understand and measure historical data that could be a valuable input for further designs and identification of potential savings of materials utilized in operation.