Automated Essay Scoring — NLP Feature Grading Model (2D)

This simulator builds a synthetic essay word by word from adjustable quality knobs — target length, sentence-length target, vocabulary diversity and spelling-error rate — then runs a real, generic feature extractor over the resulting text exactly as it would over any real submission: counting words, splitting sentences to measure average sentence length, computing the type-token ratio from the actual set of unique words used, and flagging spelling errors with a genuine dictionary lookup. Those four real numbers feed a transparent weighted-regression scoring model, and a live bar chart breaks the final predicted score down into each feature's exact contribution — the same interpretable, feature-driven grading approach visualised in three dimensions in the 3D version of this simulator, here laid flat so every step of the pipeline, from generated text to final number, is visible at once.