The motive of this study is to learn about the behavior of people on social media websites like Twitter, Facebook, etc. In this study, we focus on deepfake tweet data collected by the Twitter Platform via the Twitter API. In today’s era, the prevalence of fake things has increased a lot, which can reduce the respect of an individual and have many side effects. To understand these things, we have applied some numerical features, like “word_count”, “character_count”, “digit_count,” “average_wordlength,” “stopword_counts,” “uppercase_counts” etc., of text data. Because somewhere these numerical features represent the behavior of that text or sentence, we also apply a machine learning model such as Random Forest and calculate the performance parameters such as precision, recall, and f1-score and reach an accuracy level of 85%.