Machine Learning improvements to the accuracy of predicting Specific Language Impairment

Specific Language Impairment (SLI) has been a main obstacle of children's growth in the long-run. To diagnose this by applying Machine Learning (ML), we argue that it is crucial to understand which models should use to generate an accurate result. The key to this is the identification of positive between a large number of children, which relates to the big data that the machine has to read. In ML, the accuracy can be affected by the accidental delete of important data, failure of interpreting positive due to the random selection, imbalanced data and incorrect expression of specific features like the probability set. These problems might affect the overall accuracy in machine. Based on these issues, we gave a series of models that might be capable of solving it and train them in ML, then we can determine which model gives the best accuracy. We then consider that the big data might affect our results on accuracy, and therefore we introduced PCA. Generally speaking, we theoretically present the effectiveness of implementing different algorithms and models, which leads to the conclusion that the usefulness of application in models to the problem of diagnosing SLI.