Several volatile organic compounds (VOCs) are present in the exhaled human breath whose concentration can vary depending on the physiological changes occurring within a human being. These changes in the concentration or the occurrence of a particular VOC can be used as signature of a particular disease in a person. In the present work, a sensor has been developed to detect VOCs such as 1,4-dimethoxy-2,3-butanediol (BD), and cyclohexanone (CH), acetone, methanol and ethanol. Except for BD and CH, the rest of the VOCs are present in a healthy person in ppm levels. CH and BD have been reported to be present in the exhaled human breath of breast cancer patients in ppm levels and can be used to distinguish between a healthy person and a person with breast cancer. The selectivity of the sensor towards these two compounds in the presence of other VOCs commonly present in human breath like acetone, ethanol and methanol has been studied. The sensor has been developed using modified Quartz Tuning Forks (QTFs) with the intent of developing an array of such sensors identifying different VOCs present in a healthy human's breath. Two differently modified QTFs have been used to detect 1 ppm of 1,4-dimethoxy-2,3-butanediol and 20 ppm of cyclohexanone. Linear Discriminants Analysis (LDA) has been used to classify seven different VOCs. For this purpose, features extracted from sensor responses -shift in resonant frequency, response time and recovery time of the sensors- have been used as features in the model. Differently modified array of QTFs along with the use of LDA can be a useful pathway towards development of a QTF based sensor array for human breath analysis.