Application of partial least squares regression to relate tastiness of boiled potatoes to chemical and physical tests

Abstract

Partial Least Squares regression (PLS) was used to understand the relationship between 15 sensory attributes of potatoes boiled in their skins as dependent variables (Y), and six objective (chemical and physical) measurements as the independent variables (X). The ARC-Sensory Analysis Unit, at Irene, South Africa, conducted the study. A trained sensory panel (n=10) was used to determine the texture, aroma and flavour attributes of five potato cultivars; Mondial, BP1, Up-to-Date, Van der Plank and Caren. Four repetitions of each sample were used. The PLS regression formed three distinct groups of the cultivars: (1) Mondial, (2) BP1 and Van der Plank, which contrasted with (3) Caren and Up-to-Date. These groupings were confirmed by the culinary uses of the cultivars and their ability to retain shape after boiling. Mondial is favoured for making potato salad or boiling as it retains its shape during cooking. BP1 and Van der Plank are suitable for most uses of boiled potatoes as they retain their shape, whereas Caren and Up-to-Date do not retain their shape during cooking and are suitable for baking and frying.