Polymer modified Quartz tuning fork based sensors are a versatile replacement for the conventional sensors available in the market for the development of sensor array for electronic nose applications. However, there is not enough work towards detection of mixture of analytes using these sensors. In this work, a sensor array is developed to detect mixture of styrene, propylbenzene and acetone. Based on the working principle of the sensor, parameters like recovery time, response time, and frequency shift, among others, are utilized for data classification of the sensor response. The data collected from the sensor is used to classify the several groups using ensemble classifier, where, support vector machines, k-nearest neighbour and artificial neural network participated as the voting algorithms. Initially, the classifier was able to make prediction with approximately 93% accuracy. After modification in the prediction process where classifier was trained in a progressive manner, in which it predicted only if the data point belonged to mixture or not and subsequently, if it belonged to binary or ternary mixture. This improved the overall accuracy to 98%. The work demonstrate that the combination of QTF sensor array and a suitable classifier can be used to detect binary and ternary mixtures and therefore, can be used for several applications of electronic nose.