Regression on TOEFL11SUBSET reaches a Pearson correlation of 0.8 with proficiency scores¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2016 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Modeling TOEFL11SUBSET proficiency as regression with all features achieved a Pearson correlation of 0.8 and MAE of 0.4 in 10-fold cross validation. → Sowmya Vajjala 2016

Evidence¶

Sowmya Vajjala 2016¶

Sowmya Vajjala. (2016). Automated assessment of non-native learner essays: Investigating the role of linguistic features. International Journal of Artificial Intelligence in Education. https://arxiv.org/abs/1612.00729

q2 · i? · associational · r2

Regression experiment (SMOReg, 10-fold CV) treating the three TOEFL11 proficiency categories as an ordinal numeric scale. The all-features model obtained "a Pearson correlation of 0.8" with actual proficiency values.

"The model achieved a Pearson correlation of 0.8 and a Mean Abs olute Error (MAE) of 0.4 with actual proficiency values."

Discussion¶

Related Claims¶

• A broad linguistic feature set yields 73.2% classification accuracy for three-level proficiency scoring of TOEFL11 essays — related
• On the FCE corpus the feature set performs comparably to prior work (r = 0.64 on test), though below one coherence-feature study — related