The increasing use of social media sites in countries like India has given rise to large volumes of code-mixed data. Sentiment analysis of this data can provide integral insights into people's perspectives and opinions. Developing robust explainability techniques which explain why models make their predictions becomes essential. In this paper, we propose an adequate methodology to integrate explainable approaches into code-mixed sentiment analysis.