An Efficient Stacking Ensemble Technique for Success Prediction of Indian Ventures

India is going through a cardinal transformation right now, owing to the continuously growing home-grown startups. This project aims to use Machine Learning techniques in order to identify the factors which could potentially have an impact on the success of a startup. We have web-scraped through multiple websites on the internet to obtain a dataset with many features describing numerous Unicorn and non-Unicorn Startups in India. Techniques of Machine learning classifiers have been applied in order to predict which companies have the potential to be successful in the future. Naive Bayes Classifier, Decision Tree Classifier, Random Forest Classifier and K-Nearest Neighbor have been used. Further, a stacked ensemble model with all the above models has also been tested to provide better accuracy. The models have been evaluated through various parameters like Accuracy, F1-Scores and AUC scores. The model has been implemented through Python programming language and Random Forest has been observed to have the highest accuracy of 78.8% and an AUC score of 0.79. The stacked ensemble model provides an accuracy of 80.1%. The model was then used to predict the success rate of startups that received investment in the first season of the reality show, Shark Tank India.