We analyse and classify the sentiments of a text data constructed from movie reviews. For this, we use the kernel-based approach from quantum machine learning algorithms. In order to compose a quantum kernel, we use a circuit constructed using a combination of different Pauli rotational gates where the rotational parameter is a classical non-linear function of data points obtained from the text data. For analysing the performance of the proposed model, we analyse the quantum model using decision tree, gradient boosting classifier, and classical and quantum support vector machines. Our results show that quantum kernel model or quantum support vector machine outperforms all other algorithms used for analysis in terms of all evaluation metrics. In comparison to a classical support vector machine, the quantum support vector machine leads to significantly better results even with increased number of features or dimensions. The results clearly demonstrate increase in precision score by $9.4 \%$ using a quantum support vector machine as against a classical support vector machine if the number of features are $15$.